Radiologic Clinics of North America Body MR Imaging, Volume 41 Number 1 2003

Radiol Clin N Am 41 (2003) xi
Preface
Body MR imaging
David A. Bluemke, MD, PhD

Guest Editor
Body MR imaging applications continue to grow at

a rapid pace. In our hospital, the number of examinations performed is limited primarily by the number of
MR scanners available. Body MR imaging continues
to be one of the largest areas of growth in terms of the
percentage of examinations performed. Fellows applying for training in diagnostic imaging are highly
interested in body MR imaging facilities at various
hospitals and consider body MR imaging as a leading
reason for selecting their fellowship programs. Traditional textbooks on the subject, however, have
difficulty keeping pace with the rapidly changing
nature of imaging in this field. It is for these reasons
that this issue of the Radiologic Clinics of North
America focuses on body MR imaging applications.
Recently, outstanding advances in body MR methods have resulted in an increase in the use of MR for
patients with a variety of suspected diseases. For example, new contrast agents are now available for use in
hepatic imaging, and advanced clinical trials for intravascular contrast agents are taking place. Faster pulse
sequences that result in improved image quality with
higher resolution are available; these have helped to
control problems of motion in the chest and abdomen.
Entirely new applications that barely existed several
years ago, such as MR cholangiopancreatography,
adrenal, and breast MR imaging, are widely requested
by referring physicians. Cardiac MR imaging is of
great interest and is developing rapidly.
The complexity of these newer applications

remains a challenge for both radiologists and technologists, because comprehensive information is
usually not readily available in the literature. Increasingly, referring physicians or patients themselves are
requesting the latest MR methods and applications
from radiologists. For these reasons, an issue devoted
to body MR imaging should serve to update both the
practicing clinician and academician on a wide range
of established and novel methods. The authors chosen
for these articles are experienced in their fields
and have a broad range of expertise. This is a unique
opportunity to present an up-to-date summary of
the most widely applied applications in body
MR imaging.
I would like to thank all of the authors for their
outstanding efforts to assemble this issue of the
Radiologic Clinics of North America. In addition, I
wish to thank Barton Dudlick for the support he
provided to me as this issue was being planned and
prepared. It is my hope that readers will find this
issue a valuable resource in their current practice.
David A. Bluemke, MD, PhD
The Russell H. Morgan Department of Radiology
and Radiological Sciences
Johns Hopkins Hospital
600 North Wolfe Street, MRI 143
Baltimore, MD 21287, USA
0033-8389/03/$ see front matter D 2003, Elsevier Science (USA). All rights reserved.
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 1 2 5 - 2
Radiol Clin N Am 41 (2003) 1 15
MR physics of body MR imaging

R. Todd Constable, PhD
Department of Diagnostic Radiology, Yale University School Medical Center, 330 Cedar StreetFitkin B,
Post Office Box 208042, New Haven, CT 06520-8042, USA
Requirements and challenges of body imaging
Introduction to physics of MR
Body MR imaging has progressed significantly in

recent years due to a number of developments in the
MR pulse sequence design and acquisition techniques
[1]. This article reviews some of the challenges of MR
imaging in the body, and describes a number of
proposed solutions. Successful MR imaging in the
body requires that a number of conditions be satisfied.
These include the need for high-resolution images,
large field of view (FOV), and insensitivity to field
inhomogeneities and motion. The need to satisfy all of
these requirements make imaging of the body much
more difficult than imaging of the head or extremity,
which may require that only a small subset of these
conditions be satisfied. Motion is one of the primary
challenges in body imaging because it produces
artifacts from many sources including movement of
the cardiac wall, blood flow effects in the chambers or
vessels, respiratory effects, and peristalsis. Many
advances have been made in body imaging that
minimize these problems and satisfy the basic requirements for successful image acquisition. To understand
these methods and develop improved methods, it is
necessary to first understand the underlying physics of
MR signal behavior. This article focuses on the
physics of MR, in particular, the process of data
acquisition in MR and the influence of motion on
different acquisition strategies.
Because many of the problems encountered in

body imaging arise from motion effects and the
manifestation of these problems is highly dependent
on how the data is collected, it is important to first
discuss the acquisition of the raw data that is needed
to make an image. Although relaxation effects play a
major role in determining the contrast characteristics
of an image, they are not reviewed in this article
(other articles in this issue thoroughly describe the
contrast obtained with a wide range of different
sequences and parameters).
With regard to how the data is acquired, first
consider a hypothetical one-dimensional (1D) image
consisting of a top hat function (Fig. 1b), and the
time-domain representation of this image (see Fig. 1a).
The 1D sinc functiondefined as sin(x)/x (see Fig.
1a)is representative of the data that is collected in an
imaging experiment, either a gradient echo or a spin
echo. The data collected in an MR imaging experiment
may be thought of as composed of many of these
damped sinusoids superimposed, with a distinct sinusoid for every voxel in the image. Most of MR imaging
can be described in terms of the effects of sampling a
function such as this. This function is simply the timedomain representation (data) of the spatial frequency
domain object (image), and it is possible to move back
and forth between these representations using the
Fourier transform (FT). Note that the FT can be applied
in one dimension (as in the example in Fig. 1), two
dimensions, or three dimensions.
Much of MR imaging can be described by examination of this FT pair. The rate at which the timedomain data is sampled, and the amplitude of the
readout gradient applied during this sampling deter-
E-mail address: todd.constable@yale.edu
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 7 0 - 2
R.T. Constable / Radiol Clin N Am 41 (2003) 115
Fig. 1. One-dimensional example illustrating how ghost images are formed in the presence of an oscillating phase variation
introduced by motion of the object being imaged. The data for the true image (a) in the absence of motion produces a perfect
image (b), but is often modulated by a phase variations due to motion (c) resulting in corrupted data (e). This can also be thought
of as the true image convolved with the FT of the periodic phase variation, which is shown in (d), and which yields the same
image set containing ghosts (f).
mines the FOV and, in part, the noise present in the

image. Sampling at a higher rate increases the bandwidth and reduces the sampling time, but at the cost
of increased noise. The resolution of the image is
determined by how many points along the timedomain function are sampled. For example if the
readout gradient and sampling rate are held constant,
then doubling the number of sample points will
double the spatial resolution in the image. A good
description of sampling strategies is provided in a
book by Haacke et al [2].
If the data shown in Fig. 1a is modulated by an
oscillating phase function (see Fig. 1c), then the final
image detected (Fig. 1f) can be described as the
convolution of the FT of the oscillating function (see
Fig. 1d) and the true image (see Fig. 1b), which will
yield the true image along with ghost images (Fig. 1f)
that are periodically spaced at intervals inversely
proportional to the frequency of the oscillations. That
is, low-frequency oscillations will produce ghost
images that are close together, whereas higher-frequency oscillations will move these ghosts further
apart. Multiple components of these phase variations
will produce multiple ghosts, which may constructively or destructively interfere with one another.
Image data
In practice, an image is formed from a twodimensional (2D) or three-dimensional (3D) data
set that is comprised of many damped sinusoids in
the x-direction, y-direction, and in 3D aquisitions in
the z-direction. The data space is often referred to as
k-space because of the mathematical notation
typically used to describe points in this space. The
k-space data encodes the spatial information for the
MR signals in terms of frequency (how fast the sinusoid is oscillating) in the readout x-direction, and
in terms of phase (whether it is at a peak, trough, or
somewhere in between in the oscillatory cycle at a

particular time) in the y-direction. A spatial encoding
gradient applied in the x-direction ensures that the
frequency of the signals across the object varies
spatially in x. The k-space data is obtained line by
line, by recording data such as shown in Fig. 1a,
with different amounts of phase-encoding gradient
applied prior to the acquisition of each kx line, to
build up the entire data set (for a review of k-space
sampling, see Hennig [3]). In general, each echo
produces one line along the readout (kx) axis, and
the phase-encode (ky) data is built up point by point,
by acquiring many echoes each with a different
phase-encoding gradient applied. When no phaseencoding gradient is applied (the ky = 0 line), the
data acquired is analogous to that shown in Fig. 1a.
Motion artifacts
If motion occurs between the acquisition of the
different ky lines, then an incorrect phase will be
assigned to the ky line being acquired. In the phase
encode or y-direction, because the phase of the signal
recorded assigns position in this direction, phase errors
will result in position errors in the y-direction. Completely randomized motion adds completely randomized phase errors to each line of data, which results in a
randomized placement (or displacement) of intensities
in the phase-encode direction of the image. Periodic
motion (such as that occurring from respiration or flow
effects), on the other hand, will add cyclic phase
variations. In the phase-encode direction, phase determines position; therefore incorrect phase will lead to
intensity misplacements and hence ghosts (if the misplacements vary periodically) or intensity streaks (if
the misplacements are closer to randomly distributed).
Motion artifacts are referred to as ghosts because
they lead to additional ghostlike images of the moving structure within the imaging FOV. In most cases,
motion causes the worst artifacts to be propagated in
the y-direction due to the slow process of encoding
phase [4]. In conventional imaging sequences, a
single line of k-space data is acquired every time of
repetition (TR). Thus, to completely fill k-space,
requires Ny TR, where Ny represents the number
of phase-encode steps. This slow acquisition process
allows plenty of opportunity for motion to occur. The
frequency encoding typically takes place on the order
of less than 10 milliseconds, and thus only the most
rapid or severe motion causes artifacts in the frequency encode direction.
It is clear then that one of the biggest challenges in
body MR imaging is to collect the data without the
phase errors that arise due to motion. The next

sections examine a range of imaging sequences and
discuss how different data-collection strategies can
influence the sensitivity to motion and either avoid or
reduce the effects of motion.
Conventional image acquisition

In conventional spin echo or gradient echo
imaging, each line of data is acquired in a different
TR interval. If motion is present, then each line may be
acquired at a different position (phase) in the motion
cycle. Fig. 2a shows the magnitude of a typical
imaging data set, with damped sinusoids projecting
in all directions out from the center of k-space. If no
motion is present, then these data will appropriately
represent the object, and applying the 2D FT to the
data yields the image shown on the right of Fig. 2a. If
however, periodic motion is present, then there may be
phase variations across the data in the phase-encode
(vertical) direction, as shown in Fig. 2b. When the 2D
FT is applied to this corrupted data, motion artifacts
are clearly seen in the image to the right.
It is important to emphasize that cyclic motion
will produce discrete periodic ghosts, whereas random motion will produce more random ghosting.
There also can be more than one source of motion,
and it need not encompass the entire image. For
example, motion is observed arising from the heart
wall in Fig 3a; a streak artifact is apparent from blood
flow in a vessel in Fig. 3b; and the entire chest wall in
Fig. 3c is moving, resulting in the entire image
having multiple ghost images.
It is possible to think of these effects in terms of
each organ, vessel, or structure of the image having
its own k-space data and its own phase-modulation
function. The k-space data recorded in MR is the sum
of these separate data sets. Thus, it is difficult to
simply attempt to phase correct the complete data set
as this may eliminate motion artifacts from some
structures but may, in turn, introduce artifacts in
structures that were not moving. It is generally more
effective to devise methods that allow for the collection of the data without phase errors.
Motion artifacts: motion

compensation approaches
The earliest approaches to reducing motion involved techniques to measure motionprimarily
arising from respirationand to order the acquisition
of k-space such that an oscillatory phase pattern did
Fig. 2. Illustration of raw data (left) and the image derived from the raw data (right), by using 2D FT, analogous to the 1D
example shown in Fig. 1. In the absence of motion, the data do not have phase errors and a true image is reconstructed. (a)
Motion may introduce an oscillating phase variation, in the phase-encode (vertical ) direction, which translates into ghosts
(b, lower right ).
not arise across the data in the phase-encode direction. One of the first highly successful approaches
to this problem is called respiratory ordered phase
encoding (ROPE) (see Bailes et al [5]). This approach
makes use of respiratory bellows to measure the
cyclic nature of the patients breathing pattern before
image acquisition, and then the different phaseencode lines are collected in a temporal order
designed to create a single-phase ramp across
the data instead of an oscillatory phase variation. A
phase ramp across the data will result in a small
displacement of the object but will have little impact
on the image quality with no ghosting. Typically, the
k-space data is collected starting at the bottom of the

data set and moving up one line at a time (1, 2, 3,. . .,
Ny; where Ny = the number of phase-encode lines)
until the last line at the top is collected. In ROPE, this
order is discarded and lines 1, 10, 20, and so forth
may be collected first; followed by lines 2, 11, 21,
and so on until all of the data has been acquired. The
actual order of the lines is determined to be that
which produces a single continuous phase ramp
across the data for the given patients particular
breathing pattern. This approach is highly effective
if the patients respiratory pattern is regular and does
not change during the data-collection period. Varia-
Fig. 3. Examples of different manifestations of motion artifacts. (a) The motion artifacts are produced by the moving myocardial
wall. (b) Flow in a vessel produces a streak artifact in the phase-encode direction. (c) The entire body is moving introducing
ghosts that involve the entire FOV. In all cases, the phase-encode direction is vertical.
tions from the pattern initially sampled will lead to

more complex phase variations across the data and
result in motion artifacts. An advantage of ROPE,
however, is that it can improve image quality without
increasing the imaging time, in that a line of data is
collected with each TRas in the case of non-ROPE
conventional image-acquisition strategies. ROPE
however, will only correct for respiratory induced
ghost artifacts and does nothing to improve motion
artifacts arising from other sources such as the heart
or flow effects.
Another approach is to monitor the motion of the
chest wall or diaphragm, if respiration is the primary
concern, and collect data only when the object is in
a specific position. Although this method has the
disadvantage of dramatically extending imaging
time, when combined with fast imaging techniques
[6] it can lead to an effective means to reduce
motion artifacts.
A recent twist on this approach is called phase
ordering with automatic window selection (PAWS),
proposed by Jhooti et al [7]. In this procedure, the
motion of the diaphragm is sampled using navigator
echoes (described below) instead of respiratory bellows, and k-space data is collected continuously
throughout the motion cycle until a complete data
set is obtained that spans only a small window of the
motion cycle. The disadvantage of this approach is
the increased imaging time that results, but the
advantage is high-quality motion artifact-free images.
PAWS has recently been applied in MR coronary
angiography (MRCA) [7]. Many other approaches to
reducing motion, such as smart averaging techniques,
can be applied without the need for monitoring of the
motion [8 10].
Rather than using respiratory bellows to sample
the respiration rate, it is often more efficient to use
a navigator echo to sample the motion of a specific
region either inside or outside of the FOV to get
a more precise measure of position throughout
the motion cycle. This can be accomplished using
navigator echoes [11]. Navigator echoes are simply
projects of an image onto a single axis, and can
be obtained either just prior to the acquisition of a
k-space line or immediately after the k-space data is
acquired. Some approaches to using navigators have
examined projects of a line through the diaphragm to
measure respiratory motion as described above.
These approaches also can be used to measure the
position of the heart for cardiac imaging applications
such as MRCA [12 16] (for a review of MRCA
techniques, see Li et al [13]). Because only a projection of the object that is being imaged is required to
determine its position, a navigator may be obtained
with only a single echo with no phase-encode gradient applied. Therefore, navigators can be obtained
in as little as 8 milliseconds or less and do not
significantly impact imaging time.
Although much of this discussion has focused on
motion artifacts in the phase-encode direction, constant flow can produce displacement artifacts and
signal loss in the x-direction. These are particularly
troublesome in MRA applications, but can be minimized using gradient moment nulling methods [17
19]. In this approach, both the zero-order and firstorder gradients are nulled, such that spins moving at
constant velocity will have zero phase at the center of
the echo. This approach increases the demands on the
gradient hardware and may increase the minimum
echo time (TE) obtainable because of the additional
gradient lobes, which must be included but otherwise
do not affect the image acquisition strategy. This
approach not only reduces flow effects in vessels,
but also has been shown to improve imaging of the
spine [18,20]. Gradient moment nulling does not
abrogate the possibility of using the ROPE method,
and the two are combined easily to further reduce the
sensitivity to motion.
Two additional approaches for reducing motion
artifacts are (1) fast imaging pulse sequences to freeze
the motion, and (2) imaging acquisition strategies
that allow for the production of excellent images
from reduced data sets. These approaches are linked
in that one can collect the data faster, or one can
collect less data, which also leads to faster imaging.
The approaches are discussed separately, but the
reader should be aware that one or more of these
approaches may be applied together to reduce the
effects of motion.
Imaging faster: pulse sequences

If the image data can be obtained rapidly enough,
then there is little time for motion to occur
and therefore little opportunity for motion to corrupt
the data. By faster imaging, we imply that the data
is collected at a faster rate but that overall, a
complete symmetric data set is obtained. Fast
imaging sequences include fast spin echo (FSE),
echo planar imaging (EPI) and spiral imaging, true
fast imaging with steady-state free precession (true
FISP), and other fast gradient echo methods that are
hybrids between conventional multishot acquisitions
and single-shot EPI or spiral imaging. (Multishot
implies that more than one excitation pulseand
hence more than one TRis required to collect an
entire k-space data set, whereas single shot implies
that a single excitation pulse and a single TR

interval is used to collect the entire data set.) The
recent improvements in gradient hardware allow for
these sequences to be run on most clinical imagers.
The hardware requirements include the need for
strong gradients (on the order of 40 mT/m) and fast
gradient rise times (on the order of 200 microseconds). These capabilities allow for very short
echo times, rapid traversal of k-space, and highresolution imaging. In body imaging applications,
the gradient hardware is no longer limited by design
limitations, but rather by dB/dt safety limitations
(where dB/dt = rate of change of the gradient fields).
If the gradients are ramped too quickly they can
induce currents in the body that could stimulate the
heart or peripheral nerves. FDA guidelines limit
these ramp times to an order of magnitude below
the stimulation threshold, providing a substantial
margin of safety.
The most significant advantage of fast imaging
sequences is that they allow for complete data
acquisition in less than 30 seconds and thus are
compatible with breath-hold acquisitions, which
greatly reduces the effects of motion from breathing.
Some of these sequencessuch as single-shot EPI or
spiral imagingcan collect an image in less than
100 milliseconds, thereby freezing all but the most
rapid motion. The most commonly encountered of
these sequences are discussed in detail in the sections

that follow. The reader should be aware that there are
numerous variations on these sequences, a discussion
of which would require an entire book and not a
single article.
FSE imaging
FSE, turbo spin echo (TSE), or rapid acquisition
with relaxation enhancement (RARE) imaging, first
proposed by Hennig et al [21], acquires multiple spin
echoes following a 90 rf excitation pulse. Each of
these spin echoes has a different phase-encode gradient applied, such that multiple lines of k-space are
obtained following a single excitation pulse. The
number of echoes acquired, following each 90 rf
excitation pulse, is referred to as the echo train length
(ETL), and increases in ETL reduce imaging time in
direct proportion. For example, an acquisition with
ETL = 16, reduces imaging time compared with a
conventional spin echo acquisition (which has ETL =
1) by a factor of 16. A pulse sequence diagram for
FSE is shown in Fig. 4.
There are several potential problems introduced
by this method, one of which is blurring caused by T2
decay across the length of the echo train [22], which
can be minimized by reducing the ETL or the echo
spacing. This creates a clear trade-off between re-
Fig. 4. (a) Pulse sequence diagram for FSE imaging. The horizontal axis represents time, the vertical axis indicates when
gradients and rf pulses are on or off, as well as the polarity of the gradient applied. For the FSE sequence shown, 5 echoes are
collected (one after each 180 rf pulse), each with a different amount of phase encoding, with the colored fill of the
data acquisition windows, related to the actual lines in k-space that are collected, as shown in (b). In this example, the center of
k-space is acquired at the fifth echo, and thus the effective TE is the time between the initial 90 pulse and the fifth echo.
ducing blurring and reducing imaging time. A second

problem is that long ETLs can reduce the number of
slices that can be acquired in a given TR. Modern
magnets with good gradient hardware, however, can
usually acquire echoes rapidly enough (with only a
short delay between each echo), such that blurring
and multislice acquisitions are not significant problems. Furthermore, with the time saved by increasing
the ETL to 16 or more, it is possible to increase the
TR to accommodate more slices while still maintaining a shorter overall imaging time relative to the
conventional spin echo sequence. In its fastest form
FSE can be performed in a single-shot wherein all of
the phase-encode lines are collected in a single long
echo train, with ETL = Ny.
Contrast with FSE is determined by the echo time
at which the low-order phase-encode steps are
acquired; thus, rather than having an explicit TE
defined, FSE sequences have an effective TE
defined. The effective TE is defined as the echo time
at which the ky = 0 line is acquired. The low-order
phase-encode lines (the ky lines closest to and
including ky = 0) provide most of the image contrast
and, therefore, the echo time at which they are
acquired determines the weighting. Because FSE
acquires different lines at different echo times, it is
clear that the T2 weighting will not be the same for
all ky lines. If the high-order phase-encode linesthe
data of which provides edge informationare collected at very late echo times, the signal for these
lines will be damped due to the T2 decay, and the
images will exhibit blurring [22].
A significant difference that is immediately apparent when comparing FSE images to conventional spin
echo images is the bright fat in the T2-weighted FSE
images. The fat is bright in FSE because the multiple
180 refocusing rf pulses break up the J coupling that
is present in sequences without these pulses. This J
coupling normally leads to a loss of signal in fat in
conventional imaging sequences as magnetization is
transferred to the macromolecules of the fatty chains
[23,24]. It is possible to reduce this bright fat signal by
increasing the echo time for the first echo in the echo
traina technique aptly called DIET imaging [25,26].
By increasing the echo time for the first echo, the J
coupling is partially retained and the bright fat signal
is decreased. The later echoes still can have very short
echo spacings, as in standard FSE.
The change in J coupling with echo spacing can
also be exploited to create separate fat and water
images in a short period of time with decreased
sensitivity to field inhomogeneities [27]. By acquiring one image with very short echo spacing and a
second image with longer echo spacing but the same
effective TE, the primary signal change will be in

the fat tissue. This can be detected easily and
separate fat or water images can be formed from
this two-image set.
FSE imaging is typically used in double-echo
mode to generate both a proton density image and a
strongly T2-weighted image in a single acquisition.
With ETLs of 32 or higher, matrix sizes of 512
512 can be obtained in reasonable breath-hold times,
providing artifact-free high-resolution images. With
high-performance gradient systems, the echo spacing
can be as short as 4 milliseconds, and thus long echo
trains may be collected with minimal blurring due to
T2 decay.
The primary limitation with the FSE sequence is
the power deposition encountered when using long
trains of 180 rf pulses with short echo spacing.
Note that each echo is formed following the application of a 180 rf refocusing pulse. Thus, as the ETL
is increased and the echo spacing is decreased, the
rate of power depositionmeasured as the specific
absorption rate (SAR)increases. This can be
severe at 1.5 Tesla and increases with field strength.
Two approaches have been developed to reduce the
power deposition. The first approach, called GRASE
(gradient and spin echo) imaging, as the name
suggests, combines gradient echoes with spin echoes
[28,29]. Within an echo train, a gradient echo can be
acquired on either side of the spin echo and, if the
Fig. 5. Hyperecho pulse sequence diagram. By decreasing

the amplitude of the 180 rf pulses shown in Fig. 4, an
image can be obtained with a much lower SAR (power
deposition). In practice, the sequence of pulses 160, 140,
160, would be repeated on the other side of the second 180
pulse as 160, 140, 160, with a hyperecho forming after
the last 160 rf pulse. This last echo is called a hyperecho
because it will have the same amplitude as that which would
be obtained if all of the pulses were 180 pulses. Thus, this
sequence can provide high SNR, just like FSE, but with
lower power deposition.
phase-encode gradient is different for each echo,

then three lines of k-space may be acquired for each
spin echo, reducing the total number of spin echoes
to one third that of a standard FSE sequence, and
reducing the power by a factor of three. Just as the
total number of echoes in an echo train is flexible, so
too is the number of gradient echoes surrounding the
spin echoes.
A second approach to reducing the total rf power
deposited makes use of the concept of hyperechoes,
recently described by Hennig et al [30]. This hyperecho sequence is similar to FSE imaging, but flip
angles of less than 180 are used, thereby reducing
the total rf power (Fig. 5). By mirroring the decreased
flip angles within an echo train around a central 180
rf pulse, later echoes are formed that have the echo
amplitude that would be obtained if all pulses in the

echo train were 180 pulses. These echoes are termed
hyperechoes. The consequence of using flip angles of
less than 180 is that the contribution of stimulated
echoes increases and the image contrast reflects the
combination of stimulated and spin echoes; hence,
the images are no longer purely T2 weighted but
reflect a combination of T1 and T2 weighting.
EPI/spiral imaging
Two other approaches to fast imaging are EPI
[31,32] and spiral imaging sequences [33,34]. Both
EPI and spiral imaging can be used to generate T1weighted or T2-weighted spin echo contrast or T2*weighted gradient echo contrast. Because of the high
Fig. 6. (A) A blipped gradient echo, EPI pulse sequence. A sequence diagram demonstrates that a large Gy gradient is applied to
move to the bottom of ky-space, and then each subsequent Gy blip moves the trajectory up one line. The oscillating Gx gradient
moves the trajectory back and forth along the kx axis. The sharp transitions between positive and negative Gx place great
demands on the gradient system. This approach, however, can collect all of the k-space data following a single a excitation
pulse. The gradient echo spiral sequence shown in (B) is similar, but the smooth oscillating of the Gx and Gy gradients does not
demand the same high-performance ramping as in EPI. The spiral trajectory begins in the center of k-space (providing any TE
desired) and spirals to the outskirts of both kx and ky (a spiral-out sequence). A spiral-in sequence simply reverses the Gx and
Gy gradients shown and the minimum effective TE is much longer.
bandwidth used for data acquisition with these

sequences, the slightly different resonant frequencies
of fat and water are translated to large spatial misregistration artifacts so that these sequences are
always used with an inversion prepulse to saturate
one of these resonances. These techniques allow for
faster imaging by collecting the entire k-space data set
in a single shot (a single excitation pulse is used and
the entire data set can be collected in 65 milliseconds
or less), or in multishots wherein the data may be
acquired in 2, 4, or more shots. As with the FSE and
conventional imaging approaches discussed above, in
the limit of collecting one line of k-space per excitation pulse, a multishot EPI sequence becomes a
conventional imaging sequence with all the subsequent sensitivities to motion between shots. Multishot imaging, however, is often used in EPI or spiral
scanning to improve the image sharpness and maximize the image resolution, by reducing the amount of
T2 or T2* decay across the echo trains while increasing the matrix size, respectively. As the readout
window increases as more ky-lines are collected per
excitation pulse, blurring and off-resonance effects
increase just as in FSE. The blurring increases due to
T2* decay across the readout window, and off-resonance effects accumulate over time causing image
ghosting and distortion. A gradient echo EPI pulse
sequence diagram is shown in Fig. 6A and a gradient
echo spiral acquisition is shown in Fig. 6B.
Because of the different patterns of k-space traversal with these sequences (raster scan versus spiral),
the artifacts generated can be very different. The raster
scan approach used in EPI leads to most of the artifacts
being propagated in the y-direction, making them easy
to recognize but also quite pronounced [35]. For
example, because the odd lines of k-space are traversed
forward in time (positive x-gradient) and the even lines
are time reversed (negative x-gradient), the latter data
must be time reversed prior to application of the FT. If
field inhomogeneity effects are thought of as contributing an additional constant gradient through time, and
thus do not oscillate like the imaging gradients, then
time reversing the data introduces an even/odd phase
error across the data in the y-direction, leading to N/2
ghosts in the final image. An example of this artifact is
shown in Fig. 7. Furthermore, the accumulation of
phase errors through time leads to image distortion,
which is dependent on the local field homogeneity and
thus can be variable across the image.
If multishot EPI is used, images are at increased
susceptibility to motion effects if the object that is
being imaged is not in the same location for each shot.
In addition, N/2 ghosts may arise due to field inhomogeneities. Ghosts in a multishot EPI sequence
arising from breathing are shown in Fig. 8A, and

artifacts arising from flow effects are shown in Fig. 8B.
Spiral scanning tends to project artifacts in all
directions; thus, data errors result in blurring rather
than in discrete ghosts as seen with EPI. Note that in
spiral scanning, k-space is traversed in one or more
spirals and then the data is regridded onto the
Cartesian coordinate system such that a conventional
2D FT can be applied to reconstruct the image. Spiral
scanning has the advantage of placing fewer demands
on the gradient hardware because the gradient ramping is smooth in time. EPI, on the other hand, places
the biggest demands on the gradients, particularly
the read gradient that must ramp back and forth
between + Gmax and Gmax almost instantaneously.
Spiral imaging also has been shown to be much
less sensitive to motion than is EPI and, again,
residual artifacts that arise are spread out in all
directions, thereby decreasing their conspicuity [36].
Most spiral-scanning and EPI-scanning sequences
can collect complete images in a few seconds or less,
making breath-hold acquisitions quite acceptable to
the majority of patients.
Note also that these sequences can be adapted to
run in real time such that the operator can dynamically move the scan plane prescription in real time,
or such that very high temporal resolution may be
obtained for first-pass contrast agent studies [37,38].
A scheme for performing real-time imaging with a
four-shot spiral imaging sequence is shown in Fig. 9,
which uses the concept of view sharing to rapidly
update the image data information. Multishot spiral
scanning is particularly well suited to applications
such as this because the center of k-space is collected
Fig. 7. Ghosts from phase errors between shots in multishot

EPI. These ghosts lead to mirror images shifted by one half
or one quarter of the FOV, depending on how many shots
were used to obtain the k-space data. Shown are ghosts from
a four-shot EPI acquisition.
10
Fig. 8. Motion artifacts in multishot EPI. Much like in conventional imaging, multishot EPI is sensitive to motion that occurs
between shots and can produce multiple ghost images of the entire object as shown in (A) or they can manifest as discrete ghosts
from pulsatile flow in vessels as shown in (B).
with each shot, thereby constantly updating the

image contrast.
A final fast imaging pulse sequence to consider is
true FISP [39,40]. This is a highly motion-insensitive
sequence in that all gradients are fully balanced
within a TR such that no phase errors from motion
accumulate over time. The sequence is a fast gradient
echo sequence with very short TE and TR to minimize scan time, and it may be run in either 2D mode
as shown in Fig. 10A, or in 3D mode as shown in
Fig. 10B. This approach is used if T1-weighted
contrast is desired. Like EPI, this sequence exploits
the gradient hardware capabilities to the maximum to

minimize the TE and TR.
Imaging faster: collect less data

The final approach to improving imaging speed and
reducing imaging time to reduce sensitivity to motion
is to collect less data. There are several ways to collect
less data without dramatically decreasing image quality, resolution, or signal-to-noise ratio (SNR). The
simplest approach is to use an asymmetric FOV. This
Fig. 9. Schematic demonstrating moving window image reconstruction from a real-time four-shot spiral imaging sequence. As
each new spiral is acquired, it replaces the previous acquisition of that spiral and a new image may be reconstructed. Images are
reconstructed at the same rate that the spirals are obtained. Such an approach allows either real-time scanning, or it provides a
high temporal resolution approach for bolus tracking of contrast agents.
11
Fig. 10. The true FISP sequence is a fully balanced gradient echo imaging sequence. All gradients are balanced in this sequence
in either 2D mode (A) or in 3D mode (B). This approach is highly insensitive to motion, and provides very short TEs and TRs for
fast data acquisition.
allows for a reduction in the number of phase-encode

lines collected, without a decrease in the image resolution. Fig. 11 illustrates the concept by comparing
reduced data-sampling strategies with a conventional
256 256 matrix acquisition. If every other line is
sampled, then the FOV in the y-direction will be half
that in the x-direction, but the in-plane resolution will
remain constant. If the sampling rate is decreased
furtherfor example, to 256 96, as shown in
the example in Fig. 11then the FOV becomes too
small, and FOV wrap around artifacts (also known

as fold over artifacts or image aliasing) degrade image quality.
With the advent of multicoil imaging strategies,
however, it has become possible to use the sensitivity
profiles of the coils to unwrap these images and
thereby collect even smaller matrices with smaller
local FOVs, while still reconstructing images of
excellent quality. Two general approaches have been
developedsensitivity encoding (SENSE) [41,42]
12
Fig. 11. Imaging time may be reduced by using an asymmetric FOV. If every other line of data is acquired, thereby reducing the
data from 256 256 (a) to 256 128 (b), then the FOV in the image will be reduced by one half in the phase-encode direction,
whereas image resolution will remain constant. If the sampling rate is reduced furtherfor example, down to 256 96 (c), the
FOV will be reduced further, eventually resulting in image wrap around artifacts. These artifacts place a limit on how sparsely
the data can be sampled. If a multicoil acquisition is used, however, the sensitivity profiles of the coils may be used to unwrap
these images and produce a wrap-free image at very low sampling rates (and hence very short image acquisition times). This
multicoil approach with coil sensitivity profile data forms the basis of the SMASH and SENSE methods.
and simultaneous acquisition of spatial harmonics

(SMASH) [43,44]to unwrap these fold-over artifacts and allow for speed-up factors of two, three, or
even four.
Without multicoil acquisition strategies, it is possible to use the symmetry properties of k-space to
collect an asymmetric sample of k-space, either
collecting one half or even one quarter of k-space
as shown in Fig. 12. In general, for a real image,
points
kx and
ky are complex conjugates
of points + kx and + ky in k-space. Thus, if the data
is first-phase corrected such that the phase at kx = 0,
ky = 0 is zero, then the complete data set may be
determined with only slightly more than one quarter
of k-space sampled [45,46]. It is necessary to sample
slightly more than one quarter of k-space, such that a
small set of ky lines on either side of k-space are
obtained in order to perform the low-order phase
correction on the data. In addition to providing the
ability to unwrap these fold-over artifacts, the development of multicoil arrays has led to a significant
increase in SNR over the standard body coils for any
imaging strategy.
The asymmetric FOV sampling approach is different from the SENSE and SMASH imaging
approaches in that the sampling rate in the ky
direction is decreased (sampling every other line,
for example) in SENSE and SMASH, whereas the
sampling rate in asymmetric FOV is held constant but
the entire k space is not covered.
Recall also that many of the techniques described
above can be combined together to further reduce
imaging time while maintaining resolution and image
quality. For example, the half-acquisition single-shot
turbo spin echo [47] imaging sequence, as the name
implies, combines the TSE acquisition approach with
the half-k-space acquisition approach and takes
advantage of two of the approaches discussed in this
article to minimize imaging time and provide highquality images in the abdomen.
Overall, the combination of gradients that can
ramp up and down quickly, multicoil receivers, and
fast imaging pulse sequences, have led to highquality body imaging strategies. High-quality gradients allow for the fast traversal of k-space required
for most rapid imaging pulse sequences. Multicoil
13
Fig. 12. Another approach to reducing imaging time is to asymmetrically sample k-space. A complete image can be reconstructed
from just over one half (b), or one quarter (c) of the k-space data. The Hermitian symmetry of k-space data (kx and ky are
complex conjugates of kx and ky for real objects) can be exploited to fill in the rest of k-space. This approach may be
combined with any of the fast imaging sequences shown in the other figures to further reduce imaging time.
receivers provide excellent SNR and the ability to

collect less data without serious image wrap around
artifacts. New fast imaging pulse sequences provide
the contrast required for most clinical problems. The
best sequences can now be performed in less than
30 seconds making breath-hold acquisitions a reality
for all but the sickest patients. The combination of
these factors allows for body imaging to be performed with few artifacts and with a wide range of
choices of contrast weighting.
References
[1] Keogan MT, Edelman RR. Technological advances
in abdominal MR imaging. Radiology 2001;220:
310 20.
[2] Haacke EM, Brown RW, Thompson MR, Venkatesan R.
Magnetic resonance imaging: physical principles and
sequence design. New York (NY): Wiley Liss; 1999.
[3] Hennig J. K-space sampling strategies. Eur Radiol

1999;9:1020 31.
[4] Wood ML, Henkelman RM. MR image artifacts from
periodic motion. Med Phys 1985;12:143 51.
[5] Bailes DR, Gilderdale DJ, Bydder GM, Collins AG,
Firmin DN. Respiratory ordered phase encoding
(ROPE): a method for reducing respiratory motion
artifacts in MR imaging. J Comp Asst Tomog 1985;
9(4):835 8.
[6] Choe KA, Smith RC, Wilkens K, Constable RT. Motion artifact in T2-weighted fast spin-echo images of
the liver: effect on image contrast and reduction of
artifact using respiratory triggering in normal volunteers. J Magn Reson Imaging 1997; 7(2):298 302.
[7] Jhooti P, Gatehouse PD, Keegan J, Bunce NH, Taylor
DN, Firmin DN. Phase ordering with automatic window selections (PAWS): a novel motion resistant technique for 3D coronary imaging. Magn Reson Med
2000;43:470 80.
[8] Madore B, Henkelman RM. A new way of averaging
with applications to MRI. Med Phys 1996;23(1):
109 13.
14
[9] Xiang QS, Henkelman RM. Motion artifact reduction

with three-point ghost phase cancellation. J Magn Reson Imaging 1991;1(6):633 42.
[10] Xiang QS, Bronskill MJ, Henkelman RM. Two-point
interference method for suppression of ghost artifacts
due to motion. J Magn Reson Imaging 1993;3(6):
900 6.
[11] Ehman RL, Felmlee JP. Adaptive technique for highdefinition MR imaging of moving structures. Radiology 1989;173:255 63.
[12] Li D, Kaushikkar S, Haacke EM, Woodard PK, Dhawale PJ, Kroeker RM, et al. Coronary arteries: three
dimensional MR imaging with retrospective respiratory gating. Radiology 1996;201:857 63.
[13] Li D, Deshpande V. Magnetic resonance imaging of
coronary arteries. Top Magn Reson Imaging 2001;
12(5):337 48.
[14] Sachs TS, Meyer CH, Irarrazabal P, Hu BS, Nishimura
DG, Macovski A. The diminishing variance algorithm
for real-time reduction of motion artifacts in MRI.
Magn Reson Med 1995;34:412 22.
[15] Wang Y, Rossman PJ, Grimm RC, Riederer SJ, Ehman
RL. Navigator echo based real-time respiratory gating
and triggering for reduction of respiration effects in
three-dimensional coronary MR angiography. Radiology 1996;198(1):55 60.
[16] Wang Y, Grimm RC, Felmlee JP, Riederer SJ, Ehman
RL. Algorithms for extracting motion information
from navigator echoes. Magn Reson Med 1996;36:
117 23.
[17] Haacke EM, Lenz GW. Improving image quality in the
presence of motion by using rephasing gradients. Am J
Roentgenol 1987;148:1251 8.
[18] Hinks RS, Constable RT. Gradient moment nulling in
fast spin echo. Magn Reson Med 1994;32(6):
698 706.
[19] Katz J, Peshock RM, Malloy CR, Schaeffer S, Parkey
RW. Even-echo rephasing and constant velocity flow.
[20] Wood ML, Zur Y, Neuringer LJ. Gradient moment
nulling for steady-state free precession MR imaging
of cerebrospinal fluid. Med Phys 1991;18(5):
1038 44.
[21] Hennig J, Nauerth A, Friedburg H. RARE imaging:
a fast imaging method for clinical MR. Magn Reson
Imaging 1988;6:391 5.
[22] Constable RT, Gore JC. Loss of small objects in variable TE imaging: implications for FSE, RARE, and
EPI. Magn Reson Med 1992;28:9 24.
[23] Constable RT, Anderson AW, Zhong J, Gore JC.
Factors influencing contrast in fast spin echo
MR imaging. Magn Reson Imaging 1992;10(4):
497 511.
[24] Henkelman RM, Hardy PA, Bishop JE, Poon CS,
Plewes DB. Why fat is bright in RARE and fast spinecho imaging. J Magn Reson Imaging 1992;2(5):
533 40.
[25] Butts K, Pauly JM, Glover GH, Pelc NJ. Dual echo
DIET fast pin echo imaging. In: Proceedings of the
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
Third Scientific Meeting of the Society of Magnetic

Resonance in Medicine, Nice, France: ISMRM; 1995.
p. 651.
Stables LA, Kennan RP, Anderson AW, Constable
JC, Gore JC. Analysis of J coupling induced fat suppression in DIET imaging. J Magn Reson 1998;136:
143 51.
Constable RT, Smith RC, Gore JC. Coupled spin fast
spin echo MR imaging. J Magn Reson Imaging 1993;3:
547 52.
Mugler JP. Improved 3D GRASE imaging with the
SORT phase encoding strategy. J Magn Reson Imaging
1999;9:604 12.
Oshio K, Feinberg DA. GRASE (gradient- and spinecho) imaging: a novel fast MRI technique. Magn Reson Med 1991;20:344 9.
Hennig J, Scheffler K. Hyperechoes. Magn Reson Med
2001;46:6 12.
Mansfield P. Multi-planar image formation using NMR
spin echoes. J Phys 1977;C10:L55 8.
Mansfield P, Pykett IL. Biological and medical imaging by NMR. J Magn Reson 1978;29:355 73.
Ahn C, Kim J, Cho Z. High-speed spiral-scan echo
planar NMR imagingI. IEEE Trans Med Imaging
1986;MI-5:2 7.
Meyer CH, Hu BS, Nishimura DG, Macovski A. Fast
spiral coronary artery imaging. Magn Reson Med 1992;
28:202 13.
Feinberg DA, Oshio K. Phase errors in multi-shot
echo planar imaging. Magn Reson Med 1994;32:
535 9.
Nishimura DG, Irarrazabal P, Meyer CH. A velocity
k-space analysis of flow effects in echo-planar and
spiral imaging. Magn Reson Med 1995;33:549 56.
Kerr AB, Pauly JM, Hu BS, Li KC, Hardy CJ, Meyer
CH, et al. Real-time interactive MRI on a conventional
scanner. Magn Reson Med 1997;38:355 67.
Riederer SJ, Tasciyan T, Farzaneh F, Lee JN, Wright
RC, Herfkens RJ, et al. Technical feasibility. Magn
Reson Med 1988;8(1):1 15.
Duerk JL, Lewin JS, Wendt M, Petersilge C. Remember true FISP? A high SNR, near 1 second
imaging method for T2-like contrast in interventional MRI at 0.2T. J Magn Reson Imaging 1998;8:
203 8.
Unger EC, Cohen MS, Gatenby RA, et al. Single breathhold scans of the abdomen using FISP and FLASH at
1.5T. J Comp Asst Tomogr 1998;12:575 83.
Madore B, Pelc NJ. SMASH and SENSE: experimental and numerical comparisons. Magn Reson Med
2001;45(6):1103 11.
Pruessman KP, Weiger M, Scheidegger MB, Boesiger P.
SENSE: sensitivity encoding for fast MRI. Magn Reson
Med 1999;42:952 62.
Bydder M, Larkman DJ, Hajnal V. Generalized
SMASH imaging. Magn Reson Med 2002;47:
160 70.
Sodickson DK, Manning WJ. Simultaneous acquisition
of spatial harmonics (SMASH): ultra-fast imaging with

radiofrequency coil arrays. Magn Reson Med 1997;38:
591 603.
[45] Margosian P, Schmitt F, Purdy DE. Faster MR imaging: imaging with half the data. Health Care Instr 1986;
1:195 7.
[46] Mugler JP, Brookeman JR. Evaluation of a simple method for reconstructing asymmetrically
15
sampled echo data. J Magn Reson Imaging 1991;

1:487 91.
[47] Semelka RC, Kelekis NL, Thomasson D, Brown MA,
Laub GA. HASTE MR imaging: description of technique and preliminary results in the abdomen. J Magn
Reson Imaging 1996;6(4):698 9.
Cardiac MR imaging
Ernesto Castillo, MD, David A. Bluemke, MD, PhD*
The Russell H. Morgan Department of Radiology and Radiological Sciences, MRI Division,
The Johns Hopkins University School of Medicine, MRI-143 Nelson Basement, 600 North Wolfe Street,
Cardiac MR imaging is an effective method for

noninvasive imaging of the heart. The technology has
been limited in the past because of imaging difficulties associated with cardiac motion. In recent years,
however, cardiac MR imaging has broadened its
spectrum of applications in cardiovascular disease
with impressive advances in spatial and temporal
resolution and increased imaging speeds. This review
presents the current clinical applications of cardiac
MR imaging for evaluation of cardiac disease, to
demonstrate its application in clinical practice.
Ischemic heart disease

Myocardial viability
The concept of myocardial viability is of paramount importance for optimal clinical decision making for patients with ischemic heart disease (IHD).
Dysfunctional but viable left ventricular myocardium
(stunned or hibernating) is most likely to benefit from
revascularization, whereas left ventricle function will
not improve after revascularization (ie, coronary
bypass surgery) [1]. Revascularizing hibernating
myocardium increases global left ventricle systolic
function, reduces symptoms of congestive heart failure, and improves long-term prognosis [2]. After an
acute myocardial infarction (MI), determining infarct
size helps to stratify patient risk and determine the
extent of salvageable myocardium (Fig. 1).
E.C. is supported by a grant from the Fundacion Ramon
Areces, Madrid, Spain.
* Corresponding author.
E-mail address: dbluemke@jhmi.edu (D.A. Bluemke).
Rest imaging
The simplest approach for determining viability is
to assess end-diastolic wall thickness (EDWT) and
systolic wall thickening (SWT) on bright blood cine
images. EDWT is approximately 90% sensitive, but
only approximately 55% specific for predicting
whether there will be functional improvement postrevascularization [3,4].
Stress imaging
To improve the specificity of cine MR imaging,
dobutamine stress MR (DS-MR), using a low dose
(5 15 mg/kg/min) of dobutamine, can be performed.
The rationale for DS-MR is that contractile function of
viable tissue improves following inotropic stimulation
(contractile reserve), whereas necrotic or scarred tissue does not. Criteria for viability include EDWT less
than or equal to 5.5 mm and dobutamine-induced
SWT greater than or equal to 2 mm. In patients with
an acute or recent MI, DS-MR has a sensitivity of
approximately 91% and a specificity of approximately
70% [5]. DS-MR is therefore a better and more
accurate predictor of left ventricle functional recovery
than is preserved EDWT [3,6]. For cardiac ischemia,
DS-MR uses dobutamine doses of 10 to 40 mg/kg/
min. Stress-induced wall motion abnormality is an
early and reliable sign of myocardial ischemia that
preceeds electrocardiographic (ECG) changes and
angina [7 10]. Quantitative analysis of SWT provides better results than does qualitative analysis of
the dobutamine response (overall sensitivity of up to
91%, with 100% for three-vessel disease; and specificity of 80%) [9,10].
In a large population (208 patients) DS-MR had
better sensitivity and specificity values (approximately 86%) than did dobutamine stress echocardiog-
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 9 - 6
18
E. Castillo, D.A. Bluemke / Radiol Clin N Am 41 (2003) 1728
Fig. 1. Heterogeneity of contrast enhancement in an acute myocardial infarction in a 59-year-old man. The upper left image
shows an adequate nulling of the myocardial signal intensity at the beginning of the first-pass perfusion obtained with a
notched, interleaved, hybrid GRE EPI sequence. After 25 seconds (upper right), there is an extensive subendocardial
hypoenhancement in the territory of the left circumflex coronary artery (solid arrows), which corresponds to MO and reduces its
size at 56 seconds (lower left), due to gradual enhancement in the peripheral zone. In the image obtained at 20 minutes
postcontrast with a 3D, IR-prepared, fast GRE sequence (lower right), there is delayed enhancement around the hypointense area
of MO (arrowhead), corresponding to a transmurally extended myocardial infarction.
raphy (sensitivity 74%, specificity 70%) for significant coronary artery disease ( 50% stenosis) [7].
Currently, DS-MR is reserved mainly for patients
with inadequate stress echocardiograms due to a poor
acoustic window.
There are several limitations with DS-MR. First,
ECG changes due to ischemia cannot be detected
reliably in the high-field MR imaging environment.
Instead, visual and subjective detection of wall
motion abnormalities is performed to ensure patient
safety during stress testing. Also, wall thickening in
response to dobutamine is not entirely specific for
ischemia. This is due to through-plane motion of
the heart during the cardiac cycle, which leads to
images of different myocardial regions at end diastole versus end systole. These limitations of DS-MR
can be overcome by the use of MR imaging-tagged
cine sequences [11]. Two-dimensional (2D) tagged
MR imaging also correlates better with global
systolic ventricular function parameters than does
percentage wall thickening [12]. Clinical analysis of
tagged MR images is time consuming, however,
and therefore is used infrequently in routine clinical practice.

Contrast-enhanced imaging
Contrast-enhanced MR imaging (CE-MRI) techniques have emerged as a strong contender for
assessment of viability. Newly developed MR pulse
sequences demonstrate myocardial signal changes
following MI [13 15]. Two enhancement patterns
have been described. On the first-pass perfusion
images, there is hypoenhancement lasting approximately 20 seconds within the infarcted region. This
hypoenhanced area decreases in size over time
whereas the rest of the infarct gradually enhances.
Regions of hypoenhancement that persist for the first
minute correlate with microvascular obstruction
(MO) within the core of the infarcted area [13,16].
MO is due to damage to arterioles and capillaries with
obstruction by erythrocytes, neutrophils, and necrotic
debris [16]. The presence of MO predicts poor
myocardial functional recovery and more frequent
postinfarction cardiovascular complications [17,18].
The second enhancement pattern is observed at

10 to 30 minutes after contrast injection. This pattern is hyperenhancement in the MI area (myocardial
delayed enhancement [MDE]). Hyperenhancement is best depicted with a breath-hold segmented
inversion recovery-gradient echo (IR-GRE) technique [13,15,19]. MR images obtained 20 to
30 minutes after contrast injection reflect myocellular necrosis both after acute (1 3 days) and chronic
(8 weeks) MI, and correlate with irreversible, nonviable infarcted tissue. Thus, the concept that
bright is dead in delayed enhanced images of
the myocardium has been put forth [14,20]. Other
studies have suggested that the bright zones
overestimate the infarct zone extent by up to 10%
to 30% [21 23].
In patients with chronic MI, delayed hyperenhancement only occurs in irreversibly injured, fibrotic, or scar tissue. This allows for accurate
determination of the location and transmural extent
of healed MI [24]. The mechanism for MDE is
delayed washout and accumulation of contrast in
necrotic, nonviable myocardial tissue rich in collagen-filled fibrous scar with a larger interstitial compartment than in normal myocardium
[14,16,25]. MDE and MR imaging demonstrate that
the percentage of transmural extent of the infarcted
myocardium is inversely related to recovery of global
contractile function in both acute and chronic MI
[26,27]. In acute MI, dysfunctional nonenhancing
myocardium and enhancement of greater than 25%
of left ventricular thickness are better predictors of
long-term improvement of global contractile function
than are serum enzymes (peak creatine kinase MB) or
infarct size determined by CE-MRI itself [26]. In
chronic MI, the transmural extent of irreversibly
damaged myocardium predicts functional recovery
postrevascularization [27].
The MDE technique has been compared with
positron emission tomography (PET). There is a close
agreement between MR imaging and PET areas of
decreased flow and metabolism in severe ischemic
heart failure. MR imaging also identifies areas of
subendocardial scar tissue more frequently than does
PET, due to its higher spatial resolution [28].
Thus, high spatial resolution, high reproducibility
and predictive value, and lack of radiation exposure
make MR imaging an attractive option that may
become the standard for the assessment of myocardial
viability. Further diagnostic improvements may be
possible with the use of necrosis-specific MR contrast
agents and refinement of MR spectroscopy techniques to quantify regional chemistry and tissue
metabolism [22,29].
19
Coronary artery imaging

MR coronary angiography (MRCA) is performed
primarily at research centers, with a wide range of
reported sensitivities and specificities. Recently, the
first multicenter, prospective trial comparing MRCA
with x-ray coronary angiography has been presented
[27]. In this study, a navigator-gated, three-dimensional (3D) sequence with a T2-preparation prepulse
was used. There was a high sensitivity ( > 97%),
specificity (78% 92%), and accuracy (81% 93%)
for detection of severe coronary artery disease (>50%
stenosis on x-ray angiography) in the left main
coronary artery or triple-vessel disease. A high negative predictive value for any single diseased coronary
artery (81%) or triple-vessel disease (100%) was
shown. Thus, it may be possible to exclude the
presence of significant proximal and midcoronary
disease in patient populations with suspected multivessel disease.
Current clinical applications of MR imaging of the
coronary arteries include anomalous coronary arteries, Kawasaki disease, and assessment of bypass
graft patency.
MR coronary angiography
MRCA of native coronary arteries has been and
remains the most challenging area of cardiac MR.
There are several factors that contribute to this. The
coronary arteries are small (2 4 mm diameter) and
frequently tortuous. They have continuous inherent
cardiac (except for a brief period during middiastole)
and respiratory motion. The high signal intensity
provided by the surrounding epicardial fat is also an
added limiting factor. Both black-blood and brightblood techniques with 2D or 3D linear and nonlinear
acquisition schemes during either breath holding or
free breathing with navigators have been used with
limited success [30]. Other approaches include different techniques such as fat-suppressing prepulses,
magnetization transfer contrast prepulses, or T2 preparatory pulses [31 33].
Intravascular contrast agents under development
may allow for long time periods for image acquisition
with free-breathing high-resolution techniques.
Examples of such agents include: AMI 227 (Advanced Magnetics, Cambridge, MA), MS-325
(Angiomark; Epix Medical, Cambridge, MA), gadomer-17 (Schering AG, Berlin), or B-22956 (Gadocoletic Acid; Bracco Spa, Milan, Italy) [34 37].
These intravascular contrast agents are underongoing
investigational trials and have the potential to
improve the reliability of MRCA [37]. Submillimeter
20
in-plane spatial resolution ( 0.7 1 mm) has been

demonstrated with a free-breathing, navigator-gated,
3D, segmented, GRE acquisition scheme with a T2preparation prepulse and a short acquisition window
(70 milliseconds) [31,37]. This is still far from
the 0.3-mm to 0.5-mm isotropic resolution that is
achieved with x-ray coronary angiography, however.
aneurysms in Kawasaki disease, a rare vasculitis of

unknown origin [43]. In a study by Greil et al [43], a
free-breathing, T2-prepared, 3D, segmented, GRE
sequence with navigator gating and tracking yielded
complete agreement with x-ray coronariography in
the detection of coronary artery aneurysms, stenoses,
and occlusions (Fig. 2).
Anomalous coronary arteries
Coronary artery bypass graft patency
One of the earliest and most established indications of MRCA is the evaluation of anomalous
coronary arteries. This condition has a prevalence
of approximately 1.2%. Anomalous coronary arteries
may arise from an alternate aortic sinus of Valsalva,
from a branch of another coronary artery, or, rarely,
from the pulmonary artery [38]. These variant origins
are usually benign. If the artery travels between the
aorta and main pulmonary artery or right ventricular
outflow tract, however, there may be associated
myocardial ischemia and infarction or sudden death,
especially among adolescents and young adults [39].
X-ray coronary angiography is limited in the identification of anomalous vessels due to its projectional
nature and, therefore, is not used as a screening tool
in young adults. MRCA techniques have shown
excellent results for the identification and definition
(93% 100% of the cases) of anomalous coronary
arteries. Additionally, MRCA may classify cases that
could not be classified or were misclassified by x-ray
coronary angiography [40 42].
Differentiation between patent or occluded coronary artery bypass grafts was one of the initial
applications of MRCA. Venous bypass grafts often
have less motion, a larger lumen, and a straighter
course than do native coronary arteries. MR angiography that utilizes: black-blood and bright-blood 2D
techniques; breathholds; 3D contrast-enhancement;
and navigator gating have been shown to be useful
in demonstrating patency of bypass grafts (both
internal mammary arteries and saphenous veins)
[44,45]. Sensitivities range from 86% to 100% for
2D spin echo and GRE methods with specificities
from 59% to 96% and accuracies from 78% to 100%.
Studies with contrast-enhanced 3D MR angiography
reported greater than 93% for sensitivity, specificity,
and accuracy.
There has been limited experience using flow
measurements in bypass grafts. The determination of
the flow reserve ratio in bypass grafts may be useful for
identifying stenoses [46]. Previously, an important
limitation was the inability to evaluate all of the segments of multiple grafts within the duration of the
stress testing. Currently, however, a breath-hold,
hybrid, fast GRE echo planar imaging (EPI), phasecontrast (PC) sequence with a high temporal resolution
(23 milliseconds) now permits measurement of flow in
Kawasaki disease
MRCA has been proposed for noninvasive follow-up evaluation of children with coronary artery
Fig. 2. Large mass located ventrally of the heart with secondary compression of the right ventricular outflow tract and left
ventricle corresponding to a coronary artery aneurysm. Both black-blood (arrows, A) and cine bright-blood (arrowhead, B)
images show a small lumen.
multiple coronary artery bypass grafts both at rest

and during adenosine-induced stress test [47].
Atherosclerotic plaque
There is increasing interest in characterization and
quantification of the atherosclerotic plaque within the
coronary artery wall. The composition of the atherosclerotic plaque, rather than the degree of arterial
stenosis, appears to be a critical determinant of plaque
rupture and subsequent thrombus formation [48].
Even modestly stenotic plaques containing a large
lipid core and a thin fibrous cap are prone to rupture
(vulnerable plaque) [49]. Hence, coronary plaque
imaging could potentially differentiate clinically vulnerable from stable plaques. T2-weighted MR
imaging has primarily been used for the discrimination of fibrous and lipid components of atherosclerotic lesions, although additional pulse sequences help
to differentiate other plaque components [50 52].
Coronary artery plaque imaging is performed with
black-blood, breath-hold, T2-weighted, double-IR,
fast spin echo (FSE) sequences to obtain high-resolution images of both vessel wall (0.5 1 mm slice
thickness) and lumen (3 5 mm slice thickness)
[53,54]. The wall area and thickness of the major
epicardial coronary arteries are significantly increased
in atherosclerotic arterial segments versus normal
arteries. The main drawback of current coronary wall
imaging is the limited coverage of the coronary artery
tree. Navigator-gated 2D techniques combined with
double inversion and fat suppression as well as 3D
acquisitions have been proposed recently [36,55].
Clinical application awaits improvements in spatial
resolution for in-vivo characterization and demonstration of clinical utility.
Valvular heart disease

Echocardiography with Doppler and color-flow
mapping is primarily used in diagnosing valvular
disease. Cardiac MR is used only when other noninvasive imaging modalities fail or provide insufficient information (eg, for right ventricular valves).
Double-IR FSE sequences show valve morphology
and secondary changes of valve dysfunction (ie,
chamber enlargement, myocardial hypertrophy, poststenotic dilatation of the aorta or pulmonary artery,
and atrial or ventricular thrombus) [56]. The severity
of valvular dysfunction is assessed with cine GRE
and PC pulse sequences. Semiquantitative assessment
of stenosis or regurgitation can be obtained by
measuring the area of signal void on GRE images
21
(seen as a black jet). This is due to turbulent dephasing of spins either proximal or distal to the diseased
valve [57]. The extent of dephasing corresponds to
valvular dysfunction measured by color Doppler
echocardiography or cardiac catheterization [58
61]. The size of the MR signal void, however,
depends on acquisition parameters such as echo time
(TE), repetition time (TR), voxel size, and orientation
of the imaging plane relative to the flow jet [62,63].
Recently introduced steady-state free precession
(SSFP, commercially known as trueFISP, FIESTA, or
BalancedFFE) pulse sequences have shorter TEs and
TRs than do the conventional GRE sequences. Spin
dephasing is, therefore, much less pronounced making this an unreliable sequence for valve evaluation.
PC (velocity mapping) MR allows for quantification
and severity assessment of valve stenosis by calculating the valve orifice area and the transvalvular
pressure gradient. With this sequence, it is possible
to directly measure the peak jet velocity (Vmax, ie, in
the aorta up to 5 meters per second) at or just distal to
the valve. Emerging real-time techniques and
ongoing improvements in analysis software programs
with subsequent reduction in examination time and
data postprocessing may ease the use of MR in this
patient population, especially for pediatric patients.
Cardiomyopathies
Arrhythmogenic right ventricular dysplasia
One of the established indications for MR is the
assessment of arrhythmogenic right ventricular dysplasia (ARVD). ARVD is characterized by fatty or
fibrous infiltration with extreme thinning or thickening and wall motion abnormalities of the right
ventricle free wall [64]. These changes are the origin
for ventricular arrhythmias in adolescents or young
adults. Patients present with palpitations, syncope,
and, less commonly, sudden death. The clinical
diagnosis of ARVD is based on presence of criteria
that include structural and functional cardiac changes,
ECG abnormalities, and family history [64].
MR imaging is the most commonly used imaging
method for ARVD, due to its excellent soft tissue
contrast and the ability to depict morphology and
function (Fig. 3). To assess morphology, black-blood
double-IR FSE sequences are used. Alternatively,
gated spin echo axial images are obtained, beginning
above the pulmonary valve and extending inferiorly
down to the diaphragm. Saturation bands may be
placed over the anterior chest wall to reduce motion artifact.
22
Fig. 3. Thirty-five-year-old female athlete with positive diagnosis of ARVD. The double-IR, FSE, black-blood images without
(top) and with (bottom) spectrally selected fat suppression show high signal intensity areas within the right ventricle free wall
corresponding to fat (solid arrows), due to the signal loss with fat suppression. Note the depiction of the border between the right
ventricle myocardium and the epicardial fat (arrowheads), and the characteristic pattern of fat infiltration with irregular borders
starting from the epicardium toward the endocardium. On the cine bright-blood images (not shown), there are wall motion
abnormalities with dyskinesia in the corresponding areas of the middle and apical third of the right ventricle free wall. In
addition, there is an enlarged or dilatated right ventricle outflow tract (asterisk) in comparison with the left ventricle outflow tract
(in normal conditions both are of similar size).
Fat within the right ventricle wall is identified by

its increased signal intensity on T1-weighted images.
It can be difficult, however, to differentiate between
intramyocardial and epicardial fat [65]. Thinning
of the right ventricle free wall (normal thickness,
2.7 F 0.4 mm) is difficult to detect due to motion
artifacts and intrinsic limitations of spatial resolution
[66]. Other characteristic morphologic features are
enlargement and dilation of the right ventricle, scalloping of the right ventricle free wall, and prominent trabeculations.
Analysis of global function and regional wall
motion with cine breath-hold sequences (SSFP preferable) in the axial and short-axis planes is also performed for ARVD diagnosis. Because significant fat
infiltration of the right ventricle has been described in
greater than 50% of normal hearts in elderly patients,
functional information is critical for ARVD diagnosis
[66,67]. Areas of dyskinesis, right ventricle free wall
systolic bulging, and aneurysms are observed with MR
imaging in this condition [68].
Hypertrophic cardiomyopathy
The diagnosis of hypertrophic cardiomyopathy is
generally made by echocardiography. Diagnosis of
the hypertrophic cardiomyopathy variant confined to
the apex and involvement of the right ventricle is
easier to identify with MR than with echocardiog-
raphy, however [5,69]. Cine GRE sequences demonstrate the degree and extension of left ventricular
hypertrophy [70]. In patients with obstructing septal
hypertrophy, cine GRE sequences have been used
for semiquantitative assessment of the degree of
dynamic left ventricle outflow tract obstruction and
associated mitral regurgitation [71] (Fig. 4).
Contrast-enhanced techniques have been applied
in patients with hypertrophic obstructive cardiomyopathy for which percutaneous transluminal septal
myocardial ablation is performed. With this technique, an alcohol-induced occlusion of the first septal
branch of the left anterior descending coronary artery
is performed, creating a controlled proximal septal
acute MI. Both first-pass perfusion and delayed
contrast-enhanced MR techniques combined with
cine sequences are used to document the extent and
location of the ablation [72].
Further insights into the left ventricle mechanics
of hypertrophic cardiomyopathy have been obtained
with tagged-MR techniques, but their clinical usefulness remains to be determined [73 75]. Absolute
coronary blood flow rate per gram of myocardial
mass and the vasodilator flow reserve have been
measured with MR imaging and are significantly
lower in this patient population when compared with
healthy subjects [76]. These findings could be used to
evaluate the functional severity of the disease and
determine the prognosis.
23
Fig. 4. Hypertrophic cardiomyopathy in a 45-year-old male. The black-blood image obtained with a double-IR FSE image (left)
shows the global thickening of the left ventricle myocardium, particularly in the septum. The images obtained postcontrast (right,
top and bottom) with an IR-prepared fast GRE sequence demonstrate an irregular, diffuse pattern of enhancement mainly in the
septum. The mechanism that explains the enhancement is not well known, but is thought to be due to fibrosis.
Dilated cardiomyopathy
Dilated cardiomyopathy is characterized by ventricular enlargement with depressed contractile and
diastolic left ventricle function. Cine MR sequences
have been used for diagnosing dilated cardiomyopathy and for evaluating response to drug therapy (ie,
angiotensin-converting enzyme) [77]. The use of
gadolinium-enhanced techniques has been advocated
for the diagnosis and follow-up of dilated cardiomyopathy secondary to acute viral myocarditis. This
method is preferred over the T2 sequences used for
edema visualization [78]. Contrast-enhanced images
can detect myocardial edema early in the course of
the disease (starting on the second day). In addition,
the degree of myocardial enhancement can be correlated to clinical status and left ventricle function.
(>4 5 mm, measured on axial images) is present in

patients with constrictive pericarditis but not in
patients with restrictive cardiomyopathy [79 81].
The presence of pericardial calcification also supports
the diagnosis of a contrictive pericarditis.
Hemacromatosis
Iron deposition secondary to thalassemia major,
hemochromatosis, and other transfusion-dependent
refractory anemias causes a diffusely reduced signal
in the myocardium on cine GRE images [82]. A
significant correlation between increasing myocardial
iron (T2* below 20 milliseconds) and decline in
global left ventricle function has been found.
Pericardial disease
Restrictive cardiomyopathy
Restrictive cardiomyopathy is characterized by
diastolic dysfunction (restricted ventricular filling
and reduction in diastolic volume) of one or both
ventricles, with normal to reduced left ventricle
systolic function [79]. The condition may be idiopathic or associated with systemic disorders such as
amyloidosis, hemochromatosis, and sarcoidosis. An
established indication for cardiac MR is differentiation of restrictive cardiomyopathy from constrictive pericarditis (Fig. 5). Pericardial thickening
The most common use of MR for pericardial

disease is to differentiate between constrictive pericarditis and restrictive cardiomyopathy. Pericardial
thickness can vary in different regions of the heart;
therefore, its measurement on axial images at the levels
of the right atrium, right ventricle, and left ventricle is
recommended. The pericardium adjacent to the right
ventricle can be visualized in up to 100% of individuals; however, the pericardium along the lateral wall of
the left ventricle can be visualized in only approximately 61% of cases [83]. A thickness of more than
24
Fig. 5. Constrictive pericarditis versus restrictive cardiomyopathy. The double-IR FSE sequence (top right) shows pericardial
thickening (5 mm, solid arrow) and diastolic filling impairment with contralateral bouncing of the septum in diastole, confirmed
with the cine GRE images obtained at end diastole and end systole (bottom left and bottom right, respectively). Additional signs
of right heart failure are the enlarged hepatic veins and hepatomegaly.
4 mm is considered abnormal and a manifestation of

fibrous pericarditiseither acute or chronic, with different etiologies (infectious, uremic, connective tissue
disease, neoplasm, trauma, or cardiac surgery) [83].
Gadolinium-enhanced T1 sequences may better delineate the pericardium in cases of effusive-constrictive
pericarditis [84]. The filling dysfunction associated
with constrictive pericarditis can be seen with brightblood cine sequences performed either in the axial
plane or long-axis views. Bright-blood cine sequences
also allow for detection of enlarged hepatic veins.
Cardiac MR competes with CT as the modality of
choice in the recognition of unusual disorders such as
congenital absence of pericardium, pericardial cysts,
or tumors invading the pericardium. It is also used
when a clinically suspected pericardial effusion is not
detected by echocardiography or when specific localization or fluid characterization is desired.
Cardiac masses
Metastases to the heart are much more common
(20-fold to 40-fold) than are primary cardiac tumors.
Primary cardiac tumors are rare (0.001% 0.03% of

patients in an autopsy series) and only approximately
25% are malignant [85]. The location, size, presence
of hemorrhage, calcification, and valvular involvement are all considered in evaluating the etiology of
cardiac masses [86]. Malignant tumors often involve
the right side of the heart and pericardium with
associated pericardial effusion and extracardiac
extension (mediastinum, pleura, and chest wall)
[86]. Possible sources of diagnostic difficulty include
thrombi and the right atrial pseudomass.
Advantages of MR imaging include tissue characterization and simultaneous assessment of functional
significance and valve competence. Protocols include
the combined use of axial black-blood sequences (T1,
breath-hold, double-IR FSE with minimum TE; or, if
not available, gated, spin echo, T2, spectrally selected,
fat-suppressed FSE with 2 4 RR intervals, ETL
8 16; or breath-hold IR FSE with TE 80 100 millisecond) and axial bright-blood cine images. Gadolinium contrast is helpful in evaluating the extent of a
tumor and the site of attachment (eg, in atrial myxoma). Gadolinium techniques are identical to those used
for first-pass perfusion and MDE imaging in IHD.
25
Congenital heart disease
Summary
Cardiac MR techniques are useful in diagnosing

and characterizing complex intracardiac anatomic
details related to congenital heart diseases, such as
atrial sinus, ventricular loop, atrioventricular connection, and ventriculoarterial connections [87,88]. One
of the great strengths of cardiac MR is its ability to
image complex surgical conduits and baffles with
accurate assessment of their size and function [89].
Pulmonary to systemic flow ratios (Qp/Qs) across
shunts can be made using PC cine sequences to
calculate the magnitude and severity of a shunt or
stenosis [1,90].
There has been tremendous progress for MR

imaging depiction of cardiac morphology and function. Further advances toward achieving faster
acquisition with real-time imaging, higher resolution
for plaque imaging, and quantitative analysis are
taking place at a rapid pace.
Vascular structures
Gadolinium-enhanced, 3D, MR angiography
techniques are particularly helpful for the evaluation
of the vascular structures [91]. Using this method,
anomalies of the aorta such as vascular rings (double
aortic arch, right aortic arch, aberrant left subclavian
artery, or coarctation with or without aortic hypoplasia) are well visualized and their size can be
measured accurately. Pulmonary branch arteries and
veins are also well visualized, even in the postoperative period. Cardiac catheterization in patients
with abnormal vascular structures potentially can be
avoided using MR imaging unless knowledge of the
coronary anatomy is required or an angioplasty
is planned.
Ventricular function
Myocardial tagging techniques have provided
insight into subtle changes in the normal heterogeneous cardiac regional function of both ventricles in
infants [92,93]. In patients with congenital heart
disease, studies on single right and left ventricles
have shown the importance of interactions between
both ventricles in the cardiac contraction sequence
[89,92]. Compared with echocardiography, the right
ventricle size and function can be assessed more
accurately by MR imaging. This is most relevant in
the management of patients with disorders such as
transposition of great vessels, interventricular and
interatrial shunts, or after repair of tetralogy of
Fallot. An emerging field for MR imaging is the
examination performed in the prenatal stage, which
has provided accurate diagnosis of congenital heart
disease [94].
References
[1] Brenner LD, Caputo GR, Mostbeck G, et al. Quantification of left to right atrial shunts with velocity-encoded cine nuclear magnetic resonance imaging. J Am
Coll Cardiol 1992;20:1246.
[2] Senior R, Kaul S, Lahiri A. Myocardial viability on
echocardiography predicts long-term survival after revascularization in patients with ischemic congestive
heart failure. J Am Coll Cardiol 1999;33:1848.
[3] Baer FM, Theissen P, Schneider CA, et al. Dobutamine
magnetic resonance imaging predicts contractile recovery of chronically dysfunctional myocardium after successful revascularization. J Am Coll Cardiol 1998;
31:1040.
[4] Sechtem U, Voth E, Baer F, et al. Assessment of residual viability in patients with myocardial infarction
using magnetic resonance techniques. Int J Cardiovasc
Imaging 1993;9(Suppl 1):31 40.
[5] Devlin AM, Moore NR, Ostman-Smith I. A comparison of MRI and echocardiography in hypertrophic
cardiomyopathy. Br J Radiol 1999;72:258.
[6] Baer FM, Voth E, Schneider CA, et al. Comparison of
low-dose dobutamine-gradient-echo magnetic resonance imaging and positron emission tomography with
[18F]fluorodeoxyglucose in patients with chronic coronary artery disease. A functional and morphological
approach to the detection of residual myocardial viability. Circulation 1995;91:1006.
[7] Nagel E, Lehmkuhl HB, Bocksch W, et al. Noninvasive diagnosis of ischemia-induced wall motion abnormalities with the use of high-dose dobutamine stress
MRI: comparison with dobutamine stress echocardiography. Circulation 1999;99:763.
[8] Pennell DJ, Underwood SR, Manzara CC, et al. Magnetic resonance imaging during dobutamine stress in
coronary artery disease. Am J Cardiol 1992;70:34.
[9] van Rugge FP, van der Wall EE, de Roos A, et al.
Dobutamine stress magnetic resonance imaging for detection of coronary artery disease. J Am Coll Cardiol
1993;22:431.
[10] van Rugge FP, van der Wall EE, Spanjersberg SJ, et al.
Magnetic resonance imaging during dobutamine stress
for detection and localization of coronary artery disease. Quantitative wall motion analysis using a modification of the centerline method. Circulation 1994;
90:127.
26
[11] Geskin G, Kramer CM, Rogers WJ, et al. Quantitative

assessment of myocardial viability after infarction by
dobutamine magnetic resonance tagging. Circulation
1998;98:217.
[12] Gotte MJ, van Rossum AC, Twisk JWR, et al. Quantification of regional contractile function after infarction: strain analysis superior to wall thickening
analysis in discriminating infarct from remote myocardium. J Am Coll Cardiol 2001;37(3):808 17.
[13] Judd RM, Lugo-Olivieri CH, Arai M, et al. Physiological basis of myocardial contrast enhancement in fast
magnetic resonance images of 2-day-old reperfused
canine infarcts. Circulation 1995;92:1902.
[14] Kim RJ, Fieno DS, Parrish TB, et al. Relationship of
MRI delayed contrast enhancement to irreversible injury, infarct age, and contractile function. Circulation
1999;100:1992.
[15] Lima JA, Judd RM, Bazille A, et al. Regional heterogeneity of human myocardial infarcts demonstrated by
contrast-enhanced MRI. Potential mechanisms. Circulation 1995;92:1117.
[16] Kim RJ, Chen EL, Lima JA, et al. Myocardial GdDTPA kinetics determine MRI contrast enhancement
and reflect the extent and severity of myocardial injury
after acute reperfused infarction. Circulation 1996;
94:3318.
[17] Gerber BL, Rochitte CE, Melin JA, et al. Microvascular obstruction and left ventricular remodeling early
after acute myocardial infarction. Circulation 2000;
101(23):2734 41.
[18] Wu KC, Zerhouni EA, Judd RM, et al. Prognostic
significance of microvascular obstruction by magnetic
resonance imaging in patients with acute myocardial
infarction. Circulation 1998;97:765.
[19] Simonetti OP, Kim RJ, Fieno DS, et al. An improved
MR imaging technique for the visualization of myocardial infarction. Radiology 2001;218(1):215 23.
[20] Fieno DS, Kim RJ, Chen EL, et al. Contrast-enhanced
magnetic resonance imaging of myocardium at risk:
distinction between reversible and irreversible injury
throughout infarct healing. J Am Coll Cardiol 2000;
36(6):1985 91.
[21] Oshinski JN, Yang Z, Jones JR, et al. Imaging time
after Gd-DTPA injection is critical in using delayed
enhancement to determine infarct size accurately with
magnetic resonance imaging. Circulation 2001;
104(23):2838 42.
[22] Saeed M, Bremerich J, Wendland MF, et al. Reperfused myocardial infarction as seen with use of necrosis-specific versus standard extracellular MR
contrast media in rats. Radiology 1999;213:247.
[23] Schaefer S, Malloy CR, Katz J, et al. GadoliniumDTPA-enhanced nuclear magnetic resonance imaging
of reperfused myocardium: identification of the myocardial bed at risk. J Am Coll Cardiol 1988;12:1064.
[24] Wu E, Judd RM, Vargas JD, et al. Visualisation of
presence, location, and transmural extent of healed
Q-wave and non-Q-wave myocardial infarction. Lancet 2001;357(9249):21 8.
[25] Rehwald W.G., Fieno D.S., Chen E.L., et al. Myocardial magnetic resonance imaging contrast agent concentrations after reversible and irreversible ischemic
injury. Circulation 2002;105(2):224 9.
[26] Choi KM, Kim RJ, Gubernikoff G, et al. Transmural
extent of acute myocardial infarction predicts longterm improvement in contractile function. Circulation
2001;104(10):1101 7.
[27] Kim RJ, Wu E, Rafael A, et al. The use of contrastenhanced magnetic resonance imaging to identify reversible myocardial dysfunction. N Engl J Med
2000;343(20):1445 53.
[28] Klein C, Nekolla SG, Bengel FM, et al. Assessment
of myocardial viability with contrast-enhanced magnetic resonance imaging: comparison with positron
emission tomography. Circulation 2002;105(2):162 7.
[29] Bottomley PA, Weiss RG. Noninvasive localized MR
quantification of creatine kinase metabolites in normal
and infarcted canine myocardium. Radiology 2001;
219(2):411 8.
[30] Wielopolski PA, van Geuns RJ, de Feyter PJ, et al.
Coronary arteries. Eur Radiol 2000;10(1):12 35.
[31] Botnar RM, Stuber M, Danias PG, et al. Improved
coronary artery definition with T2-weighted, freebreathing, three-dimensional MRA coronary. Circulation 1999;99:3139.
[32] Edelman RR, Manning WJ, Burstein D, et al. Coronary
arteries: breath-hold MR angiography. Radiology
1991;181:641.
[33] Li D, Paschal CB, Haacke EM, et al. Coronary arteries:
three-dimensional MR imaging with fat saturation and
magnetization transfer contrast. Radiology 1993;
187:401.
[34] Hofman MB, Henson RE, Kovacs SJ, et al. Blood pool
agent strongly improves 3D magnetic resonance coronary angiography using an inversion pre-pulse. Magn
Reson Med 1999;41:360.
[35] Li D, Dolan RP, Walovitch RC, et al. Three-dimensional MRI of coronary arteries using an intravascular
contrast agent. Magn Reson Med 1998;39:1014.
[36] Stuber M, Botnar RM, Kissinger KV, et al. Freebreathing black-blood coronary MR angiography: initial results. Radiology 2001;219(1):278 83.
[37] Stuber M, Botnar RM, Danias PG, et al. Contrast
agent-enhanced, free-breathing, three-dimensional coronary magnetic resonance angiography. J Magn Reson
Imaging 1999;10:790.
[38] Engel HJ, Torres C, Page HLJ. Major variations in
anatomical origin of the coronary arteries: angiographic observations in 4,250 patients without associated congenital heart disease. Catheter Cardiovasc
Diagn 1975;1:157.
[39] Kragel AH, Roberts WC. Anomalous origin of either
the right or left main coronary artery from the aorta
with subsequent coursing between aorta and pulmonary trunk: analysis of 32 necropsy cases. Am J Cardiol 1988;62:771.
[40] McConnell MV, Ganz P, Selwyn AP, et al. Identification of anomalous coronary arteries and their anatomic
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
course by magnetic resonance coronary angiography.

Circulation 1995;92:3158.
Post JC, van Rossum AC, Bronzwaer JG, et al. Magnetic resonance angiography of anomalous coronary
arteries. A new gold standard for delineating the proximal course? Circulation 1995;92:3163.
Taylor AM, Thorne SA, Rubens MB, et al. Coronary
artery imaging in grown up congenital heart disease:
complementary role of magnetic resonance and x-ray
coronary angiography. Circulation 2000;101(14):
1670 8.
Greil GF, Stuber M, Botnar RM, et al. Coronary magnetic resonance angiography in adolescents and young
adults with kawasaki disease. Circulation 2002;105(8):
908 11.
Vrachliotis TG, Bis KG, Aliabadi D, et al. Contrastenhanced breath-hold MR angiography for evaluating
patency of coronary artery bypass grafts. AJR Am J
Roentgenol 1997;168:1073.
Wintersperger BJ, Engelmann MG, von Smekal A, et
al. Patency of coronary bypass grafts: assessment with
breath-hold contrast-enhanced MR angiographyvalue of a non-electrocardiographically triggered technique. Radiology 1998;208:345.
Sakuma H, Globits S, OSullivan M, et al. Breath-hold
MR measurements of blood flow velocity in internal
mammary arteries and coronary artery bypass grafts.
J Magn Reson Imaging 1996;6:219.
Langerak SE, Kunz P, Vliegen HW, et al. Improved
MR flow mapping in coronary artery bypass grafts
during adenosine-induced stress. Radiology 2001;
218(2):540 7.
Fernandez-Ortiz A, Badimon JJ, Falk E, et al. Characterization of the relative thrombogenicity of atherosclerotic plaque components: implications for
consequences of plaque rupture. J Am Coll Cardiol
1994;23:1562.
Falk E, Shah PK, Fuster V. Coronary plaque disruption. Circulation 1995;92:657.
Toussaint JF, Southern JF, Fuster V, et al. T2-weighted
contrast for NMR characterization of human atherosclerosis. Arterioscler Thromb Vasc Biol 1995;15:1533.
Toussaint JF, LaMuraglia GM, Southern JF, et al. Magnetic resonance images lipid, fibrous, calcified, hemorrhagic, and thrombotic components of human
atherosclerosis in vivo. Circulation 1996;94:932.
Worthley SG, Helft G, Fuster V, et al. Noninvasive in
vivo magnetic resonance imaging of experimental coronary artery lesions in a porcine model. Circulation
2000;101(25):2956 61.
Botnar RM, Stuber M, Kissinger KV, et al. Noninvasive
coronary vessel wall and plaque imaging with magnetic
resonance imaging. Circulation 2000;102(21):2582 7.
Fayad ZA, Fuster V, Fallon JT, et al. Noninvasive in
vivo human coronary artery lumen and wall imaging
using black-blood magnetic resonance imaging. Circulation 2000;102(5):506 10.
Botnar RM, Kim WY, Bornert P, et al. 3D coronary
vessel wall imaging utilizing a local inversion tech-
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
27
nique with spiral image acquisition. Magn Reson

Med 2001;46(5):848 54.
Arai AE, Epstein FH, Bove KE, et al. Visualization of
aortic valve leaflets using black blood MRI. J Magn
Reson Imaging 1999;10:771.
de Roos A, Reichek N, Axel L, et al. Cine MR imaging
in aortic stenosis. J Comput Assist Tomogr 1989;
13:421.
Globits S, Frank H, Mayr H, et al. Quantitative assessment of aortic regurgitation by magnetic resonance
imaging. Eur Heart J 1992;13:78.
Hundley WG, Li HF, Willard JE, et al. Magnetic resonance imaging assessment of the severity of mitral
regurgitation. Comparison with invasive techniques.
Kilner PJ, Manzara CC, Mohiaddin RH, et al. Magnetic resonance jet velocity mapping in mitral and
aortic valve stenosis. Circulation 1993;87:1239.
Yoshida K, Yoshikawa J, Hozumi T, et al. Assessment of aortic regurgitation by the acceleration flow
signal void proximal to the leaking orifice in cinemagnetic resonance imaging. Circulation 1991;
83:1951.
Bryant DJ, Payne JA, Firmin DN, et al. Measurement
of flow with NMR imaging using a gradient pulse and
phase difference technique. J Comput Assist Tomogr
1984;8:588.
Suzuki J, Caputo GR, Kondo C, et al. Cine MR imaging of valvular heart disease: display and imaging parameters affect the size of the signal void caused by
valvular regurgitation. AJR Am J Roentgenol 1990;
155:723.
Fontaine G, Fontaliran F, Hebert JL, et al. Arrhythmogenic right ventricular dysplasia. Annu Rev Med
1999;50:17 35.
Blake LM, Scheinman MM, Higgins CB. MR features
of arrhythmogenic right ventricular dysplasia. AJR Am
J Roentgenol 1994;162:809.
Burke AP, Farb A, Tashko G, et al. Arrhythmogenic
right ventricular cardiomyopathy and fatty replacement
of the right ventricular myocardium: are they different
diseases? Circulation 1998;97:1571.
Shirani J, Berezowski K, Roberts WC. Quantitative
measurement of normal and excessive (cor adiposum)
subepicardial adipose tissue, its clinical significance,
and its effect on electrocardiographic QRS voltage.
Am J Cardiol 1995;76:414.
Ricci C, Longo R, Pagnan L, et al. Magnetic resonance
imaging in right ventricular dysplasia. Am J Cardiol
1992;70:1589.
Suzuki J, Shimamoto R, Nishikawa J, et al. Morphological onset and early diagnosis in apical hypertrophic
cardiomyopathy: a long term analysis with nuclear
magnetic resonance imaging. J Am Coll Cardiol
1999;33:146.
Pons-Llado G, Carreras F, Borras X, et al. Comparison
of morphologic assessment of hypertrophic cardiomyopathy by magnetic resonance versus echocardiographic imaging. Am J Cardiol 1997;79:1651.
28
[71] Pattynama PM, Lamb HJ, Van DV, et al. Left ventricular measurements with cine and spin-echo MR imaging: a study of reproducibility with variance
component analysis. Radiology 1993;187:261.
[72] Schulz-Menger J, Strohm O, Waigand J, et al. The value of magnetic resonance imaging of the left ventricular
outflow tract in patients with hypertrophic obstructive
cardiomyopathy after septal artery embolization. Circulation 2000;101(15):1764 6.
[73] Dong SJ, MacGregor JH, Crawley AP, et al. Left ventricular wall thickness and regional systolic function in
patients with hypertrophic cardiomyopathy. A threedimensional tagged magnetic resonance imaging study.
[74] Kramer CM, Reichek N, Ferrari VA, et al. Regional
heterogeneity of function in hypertrophic cardiomyopathy. Circulation 1994;90:186.
[75] Maier SE, Fischer SE, McKinnon GC, et al. Evaluation
of left ventricular segmental wall motion in hypertrophic cardiomyopathy with myocardial tagging. Circulation 1992;86:1919.
[76] Kawada N, Sakuma H, Yamakado T, et al. Hypertrophic cardiomyopathy: MR measurement of coronary
blood flow and vasodilator flow reserve in patients
and healthy subjects. Radiology 1999;211:129.
[77] Doherty NE, Seelos KC, Suzuki J, et al. Application of
cine nuclear magnetic resonance imaging for sequential evaluation of response to angiotensin-converting
enzyme inhibitor therapy in dilated cardiomyopathy.
J Am Coll Cardiol 1992;19:1294.
[78] Friedrich MG, Strohm O, Schulz-Menger J, et al. Contrast media-enhanced magnetic resonance imaging visualizes myocardial changes in the course of viral
myocarditis. Circulation 1998;97:1802.
[79] Sechtem U, Higgins CB, Sommerhoff BA, et al. Magnetic resonance imaging of restrictive cardiomyopathy.
Am J Cardiol 1987;59:480.
[80] Masui T, Finck S, Higgins CB. Constrictive pericarditis and restrictive cardiomyopathy: evaluation with MR
imaging. Radiology 1992;182:369.
[81] Soulen RL. Magnetic resonance imaging of great vessel, myocardial, and pericardial disease. Circulation
1991;84:I311.
[82] Blankenberg F, Eisenberg S, Scheinman MN, et al. Use
[83]
[84]
[85]
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
of cine gradient echo (GRE) MR in the imaging of

cardiac hemochromatosis. J Comput Assist Tomogr
1994;18:136.
Sechtem U, Pflugfelder PW, Cassidy MM, et al. Mitral
or aortic regurgitation: quantification of regurgitant
volumes with cine MR imaging. Radiology 1988;
167:425.
Watanabe A, Hara Y, Hamada M, et al. A case of
effusive-constructive pericarditis: an efficacy of GDDTPA enhanced magnetic resonance imaging to detect
a pericardial thickening. Magn Reson Imaging 1998;
16:347.
Burke A, Virmani R. Tumours of the heart and
great vessels. In: Atlas of tumor pathology, fasc 16,
ser 3. Washington (DC): Armed Forces Institute of
Pathology; 1996.
Araoz PA, Eklund HE, Welch TJ, et al. CT and MR
imaging of primary cardiac malignancies. Radiographics 1999;19:1421.
Boxt LM. MR imaging of congenital heart disease.
Magn Reson Imaging Clin N Am 1996;4:327.
Ho VB, Kinney JB, Sahn DJ. Contributions of newer
MR imaging strategies for congenital heart disease.
Radiographics 1996;16:43.
Fogel MA, Hubbard A, Weinberg PM. A simplified
approach for assessment of intracardiac baffles
and extracardiac conduits in congenital heart surgery
with two- and three-dimensional magnetic resonance
imaging. Am Heart J 2001;142(6):1028 36.
Mohiaddin RH, Pennell DJ. MR blood flow measurement. Clinical application in the heart and circulation.
Cardiol Clin 1998;16:161.
Hartnell GG, Meier RA. MR angiography of congenital heart disease in adults. Radiographics 1995;15:781.
Fogel MA. Assessment of cardiac function by magnetic resonance imaging. Pediatr Cardiol 2000;
21(1):59 69.
Klein SS, Graham TPJ, Lorenz CH. Noninvasive delineation of normal right ventricular contractile motion
with magnetic resonance imaging myocardial tagging.
Ann Biomed Eng 1998;26:756.
Hubbard AM, States LJ. Fetal magnetic resonance imaging. Top Magn Reson Imaging 2001;12(2):93 103.
MR angiography of the chest

Florian M. Vogt, MD*, Mathias Goyen, MD, Jorg F. Debatin, MD, MBA
Department of Diagnostic and Interventional Radiology, University Hospital Essen, Hufelandstrasse 55, 45122 Essen, Germany
Within less than a decade following its initial

description by Prince et al in 1993 [1], three-dimensional contrast-enhanced MR angiography (MRA)
has been firmly established as an accurate noninvasive alternative for the diagnostic assessment of
almost all vascular territories including the intrathoracic vessels. The technique combines the intravenous administration of paramagnetic contrast
agents with the ultrafast acquisition of T1-weighted
three-dimensional gradient echo data sets [1]. Contrast-enhanced three-dimensional MRA exploits the
contrast-induced T1 shortening effects during the
intra-arterial phase of the contrast agent. The availability of high-performance gradient systems permits
the acquisition of complex three-dimensional data
sets within the confines of a comfortable breathhold
interval of under 30 seconds.
Contrast-enhanced three-dimensional MRA is
able to overcome known limitations of conventional
black blood and bright blood MRA techniques
including cardiac and respiratory pulsation artifacts,
poor signal-to-noise, susceptibility effects, and artifacts at air-tissue interfaces. Image quality is no
longer related to flow or saturation [2,3]. Cardiac or
respiratory gating is no longer necessary.
This article describes existing state-of-the-art contrast-enhanced three-dimensional MRA techniques
for the assessment of the intrathoracic arterial and
venous systems. Technique-related aspects are highlighted and the existing clinical experiences are
summarized. Finally, developing techniques, which
E-mail address: florian.vogt@uni-essen.de (F.M. Vogt).
are likely to enhance further the impact of thoracic

MRA in the future, are discussed.
Technical considerations
Contrast-enhanced three-dimensional MRA of the
thoracic vasculature offers several advantages over
conventional MRA techniques, including shorter
acquisition times and high spatial resolution in conjunction with high signal- and contrast-to-noise. In
many centers, the availability of contrast-enhanced
three-dimensional MRA has profoundly impacted
diagnostic strategies for exploring the thoracic vascular system [4 6]. Requiring only a peripheral
intravenous catheter and administration of contrast
agents characterized by an excellent safety profile
[7,8], three-dimensional MRA techniques has vastly
lowered the threshold for assessing the arterial and
venous morphology of the thoracic vessels.
Paramagnetic contrast agents are pivotal for displaying the vascular system with fast three-dimensional gradient echo sequences. Without the presence
of paramagnetic contrast, these sequences, characterized by very short repetition and echo times, render
nondiagnostic images void of any intravascular signal.
The presence of paramagnetic contrast in the vascular
system over the length of the data acquisition period is
crucial for successful contrast-enhanced three-dimensional MRA.
Paramagnetic contrast shortens the T1 relaxation
time of blood. Gadolinium (Gd), the most commonly
used paramagnetic substance, has a high relaxivity
and a favorable safety profile when bound to a
chelate. During the short intravascular phase the
intravenously injected T1 shortening contrast agent
provides signal in the arterial and venous systems,
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 1 - 1
30
F.M. Vogt et al / Radiol Clin N Am 41 (2003) 2941
elevating the vessel to background contrast-to-noise

ratio and eliminating flow artifacts. The signal of
flowing blood is no longer flow-dependent. Flowinduced artifacts seen with noncontrast time-of-flight
or phase-contrast MRA techniques are largely eliminated, and images can be collected in the plane
coinciding with the course of the vessels of interest.
This allows coronal coverage of large vascular territories in short imaging times and generates images
that are similar in appearance to conventional catheter-based radiograph angiography [9].
Although similar in principle to spiral CT angiography, contrast-enhanced MRA holds considerable
advantages. Beyond the absence of ionizing radiation
and the ability to depict large vascular territories in
three-dimensional imaging volumes, harmful side
effects of the paramagnetic contrast agents used for
the MRA examination are considerably less frequent
and less severe than those associated with the iodinated contrast used in CT angiography. Paramagnetic
contrast agents are nonnephrotoxic and have a low
incidence of anaphylactoid reactions [8]. They are
safe for use in patients with renal insufficiency and in
patients with a history of allergic reactions to iodinated contrast media. Finally, on contrast-enhanced
three-dimensional MRA images only the contrastfilled vessels are displayed. In addition, bones and
calcium remain dark facilitating interpretation and
subsequent postprocessing of the underlying threedimensional data sets.
Contrast-enhanced three-dimensional MRA has
been shown to be useful for the depiction of the
supra-aortic arteries, the thoracic and abdominal aorta,
and its major branch vessels. In its initial implementation, lengthy imaging times ranging between 3 and 5
minutes precluded data acquisition during a breathhold
[10]. Ensuing respiratory motion artifacts considerably
degraded image quality. Breathhold data acquisition
did become possible with the use of improved gradient
systems permitting considerable reductions in the
minimum repetition (TR) and echo times (TE)
[9,11]. The implementation of fast three-dimensional
gradient echo sequences on high-performance systems
permits the acquisition of complex three-dimensional
data sets within the confines of a comfortable breathhold in as little as 5 to 20 seconds [12]. This dramatic
reduction of scan time has even allowed the collection
of temporally resolved three-dimensional data sets
[13]. The transit of the paramagnetic contrast agent
through the vascular system can be depicted. Ultrafast
three-dimensional data acquisitions in conjunction
with fast table feeds also permit chasing the contrast
bolus through several vascular territories extending all
the way to whole body MRA [14,15].
Paramagnetic contrast
There are several paramagnetic contrast agents
available today. Currently, only extracellular nonbinding Gd chelates with a concentration of 0.5 mol/L
have regulatory approval for use in humans [16]. In
the United States, no agent is currently approved for
MRA by the Food and Drug Administration and any
such use constitutes off-label use of an approved drug.
Several paramagnetic intravascular Gd-based agents
and superparamagnetic compounds are currently
undergoing clinical and preclinical testing [17 20].
With higher infusion rates the local concentration
of the contrast in the vessel of interest is higher, but
because of a faster venous return there is a shorter
arterial-venous time window for imaging. Contrast
compounds with higher concentration formulations,
such as Gadovist 1 mol/L (Schering, Berlin, Germany), might be advantageous when slow infusion
protocols are used. Preliminary results comparing
a 0.5 mol/L contrast agent (Magnevist, Schering,
Berlin, Germany) with Gadovist 1 mol/L in pelvic
MRA are very promising with regard to arterial
enhancement [21].
Some newer agents, such as Multihance (Bracco,
Milan, Italy), have higher relaxivity because of a
transient binding to albumin and different routes of
excretion but are still distributed into the extracellular
space. Blood pool or intravascular contrast agents are
sufficiently large or bind to large molecules when
injected. This prevents them from leaking out of the
capillaries and confines them to the intravascular
compartment for extended periods of time. The major
disadvantage of intravascular contrast agents is the
early venous enhancement leaving a short window
for arterial imaging. This represents a particular
problem in the lower legs, where venous overlap
can seriously impair the ability to assess the arteries.
These agents are likely to play a dominant role in the
assessment of the coronary arteries [22,23].
In view of the rapid progress of MRA techniques
using extracellular agents, the future of intravascular
contrast agents for morphologic imaging of the arterial vascular tree (with the exception of the coronary
arteries) remains uncertain.
The paramagnetic agent is generally administered
by an intravenous catheter placed into an antecubital
vein. To achieve maximal image quality, the presence
of the intravenously administered contrast bolus in
the vascular territory under consideration must coincide with the data acquisition period. Several manual
and automated techniques are available to ensure
proper timing of the scan delay and are discussed
later. Maximal contrast concentration in the vessel of
interest should be achieved during the acquisition of

the central, contrast-determining portion of k-space.
Poor timing of the contrast application affects
image quality in different ways: venous overlap,
ringing artifacts, or insufficient signal within the
vessels of interest can occur [24,25]. To ensure the
exact synchronization of arterial enhancement with
the acquisition of arterial-phase Gd-enhanced threedimensional MRA, different strategies have been
developed to optimize timing including the use of
a test bolus, fluoroscopic triggering, and an automated detection system, which triggers the initiation
of scanning when contrast material is detected in a
predefined area of interest. Optimal timing of contrast administration results in three-dimensional
MRA data sets void of ringing artifacts and venous
overlap [26].
Using a fixed scan delay is the simplest means of
applying the paramagnetic contrast agent and works
in most instances for MRA of the thoracic arteries
[27]. The contrast bolus should commence 10 seconds before the beginning of data acquisition and
encompass the entire scan time. Age, body weight,
and heart rate are some of the parameters that make it
more difficult to predict the circulation time. Test
bolus timing techniques are established and simple to
use. Following the injection of a 1- to 2-mL test bolus
of paramagnetic contrast, images of the target vessel
are collected using a rapid two-dimensional gradientecho pulse sequence, preferably in the axial plane,
with a temporal resolution of one image every 1 to
2 seconds. Arrival of contrast agent is detected by a
transient signal increase, hence the appropriate time
delay to align the central k-space with arterial
enhancement can be calculated [9,28]. For centrically
encoded sequences, the delay of data acquisition
corresponds to the time the test bolus needs to reach
the target volume. Recent investigations reveal that
examinations after bolus testing showed significantly
superior signal-to-noise ratio (SNR) as compared
with examinations without a test bolus [29].
Other bolus-timing techniques are based on realtime triggering. There are two similar approaches.
Line scanning measures the signal intensity within an
operator-defined monitoring volume and subsequently initiates data acquisition following bolus
arrival automatically [30]. Hand fluoroscopic imaging uses visual cues identifying the contrast agent to
be present in the vessels under investigation. The
three-dimensional MRA sequence is subsequently
triggered manually to capture peak contrast. These
timing methods generally are based on the use of
centric view ordering of k-space for the three-dimensional acquisition. The fluoroscopic-triggered three-
31
dimensional elliptical centric view ordering technique

has shown a reliability exceeding 90% in detecting
the bolus arrival [31,32].
Faster gradient sets now permit for the acquisition
of time-resolved three-dimensional MRA. To enhance the speed of data collection further these
techniques use a host of methods including variable-rate k-space sampling, view sharing, temporal
interpolation, and zero filling. The method ensures
the acquisition of data and reconstruction of images at
time points before, during, and after the contrast
agent passes through the vessels of interest. The need
for timing the contrast bolus is eliminated [13,33,34].
Time-resolved three-dimensional MRA is highly
robust because it is virtually operator independent.
In most applications the required high temporal
resolution can be achieved only at the price of
compromising spatial resolution. The application of
sensitivity encoding and simultaneous magnetization
of spatial harmonics parallel imaging techniques is
likely to provide both high temporal and spatial
resolution [35].
Intravenously administered Gd doses range
between 0.05 and 0.3 mmol/kg body weight depending on the vascular territory under investigation.
Doses between 0.1 and 0.2 mmol/kg body weight
Gd-based contrast agent were reported to be sufficient
for most single-station MRA examinations [7,16].
Even lower dosing (0.075 mmol/kg) can be used in
time-resolved MRA [36]. The resulting injection
volume of 15 to 40 mL should be administered at a
rate between 1.5 and 2.5 mL/second. The role of
injection rate on image quality is still being determined. If the injection rate is too fast, k-space
modulation occurs, degrading image quality and
increasing ringing artifacts [25]. Carroll et al [24]
determined that lowering the injection rate reduces
changes in signal versus time and results in less
blurring without a significant loss of vessel contrast.
The actual flow rate should be adjusted to ensure
injection of the entire contrast volume in a period not
exceeding the acquisition time. Regardless of the
timing regimen or imaging technique, use of an
automated injector facilitates contrast timing and
delivery because it allows precise infusion using
predefined weight-adjusted rates and volumes.
For imaging the thoracic veins, diluted contrast
(1:10 to 20) should be injected directly into the
affected side. For a comprehensive display of the
entire venous system, both arms can be injected
simultaneously. An alternative method for central
venous imaging is to administer a diluted dose of
Gd-chelate contrast media directly into the vein of
interest [37,38].
32
Pulse sequence considerations

The pulse sequence design for contrast-enhanced
MRA is based on a three-dimensional Fourier transform gradient recalled echo (GRE) sequence using
rapid radiofrequency pulsing. Optimization of MRA is
related to rapid data collection. The fastest possible
three-dimensional imaging sequence, typically a fast
three-dimensional gradient echo pulse, should be used.
To achieve maximal T1 weighting, spoiled sequences
should be used. Spoiling is useful because it destroys
the residual magnetization after each echo and magnifies the effect of T1 relaxation agents. Repetition and
echo times should be as short as possible. The newest
generation of commercially available 1.5-T MR
imaging scanners provides three-dimensional gradient
echo pulse sequences with minimal repetition times of
less than 2 milliseconds and minimal echo times of less
than 1 millisecond. A flip angle ranging between 10
and 25 degrees provides adequate suppression of the
surrounding tissues and has been shown to render
excellent image quality: the longer the repetition time,
the lower the flip angle.
Thoracic contrast-enhanced three-dimensional
MRA data need to be collected within a breathhold
to minimize the artifacts related to respiratory motion.
The partition dimensions (ie, partition thickness,
matrix size, and field-of-view) should be prescribed
to achieve the smallest possible voxel size to allow
for sufficient spatial coverage of the target vessel.
Section thickness should be adjusted to be between
1.5 and 2.5 mm to ensure full coverage of the

vascular system under consideration and still permit
multiplanar reformations. To ensure full coverage,
between 48 and 64 thin contiguous sections need to
be collected.
Imaging time can be decreased further using
partial Fourier imaging, decreased number of partitions, decreased phase-encoding steps, or a rectangular field of view. Widening bandwidth also makes for
faster scanning, although this causes a reduction in
SNR. These improvements in image speed can be
used to improve image resolution (ie, decrease voxel
size). Zero-filling, although not improving the true
spatial resolution, generates better reformations and
maximum intensity projections and reduces partial
volume averaging errors. Sensitivity encoding and
simultaneous magnetization of spatial harmonics
imaging are newly developed techniques that enable
further reductions in scan time, while maintaining
spatial resolution (Fig. 1). The accompanying 30% to
55% reduction in SNR does not seem to affect image
quality adversely because of the high concentration of
the contrast compound in the vasculature during data
collection [39,40].
It is well recognized that, to ensure image quality,
the contrast must be near or at the peak concentration
in the vasculature of interest when acquiring the
central phase-encoding views of the MRA. In principle there are two different methods for k-space
phase ordering: conventional sequential phase ordering where the center k-space views are collected in
Fig. 1. Normal contrast-enhanced MRA of the thoracic aorta in a 28-year-old healthy volunteer acquired in the coronal plane. (A)
Maximum intensity projection (MIP) reformation of a coronal three-dimensional MRA data set using a standard phased array
coil. (B) Coronal MIP image achieved with twofold simultaneous magnetization of spatial harmonics reduction using a new sixelement cardiac array. (C) Coronal MIP image with twofold sensitivity encoding reduction. A spatial resolution of 1.7 0.8
1.8 mm3 was achieved with decrease of scan time almost to 50% in Fig. 1B and C. Despite the intrinsic drawback in signal-tonoise ratio, the coronal MIP is of nearly identical diagnostic value.
the middle of the imaging period, and centric phase

ordering that enables the acquisition of the crucial
image contrast data at the beginning of the sequence
[9]. View order with elliptic centric phase encoding
has been shown to be centric in both phase-encoding
directions, which provides minimal sensitivity to
motion artifact caused by loss of breathhold
[32,41]. To facilitate the timing of the contrast bolus
most manufacturers now provide centrically encoded
sequences for the performance of MRA.
Other techniques
The recently developed steady-state free precession (SSFP) gradient echo sequence (ie, TrueFISP,
FIESTA, balanced fast field echo) is characterized by
balanced gradients in all three directions, ensuring
maximum recovery of the transverse magnetization
[42,43]. At the end of the TR, the transverse magnetization is refocused, and the next excitation can be
started without further preparation. The image contrast with SSFP is determined by favorable T2*-T1
properties that are nearly independent from blood
flow [44]. The main advantage of this sequence
33
relates to its high signal of fluids, which results in

high contrast delineating vessels as bright structures,
while at the same time delineating all other morphology. The use of short TRs is mandatory to avoid
T2* effects and can only be achieved using the most
powerful gradient systems. The short TR and echo
time reduce susceptibility artifacts and lead to
extremely short data acquisition times. Reductions
in acquisition time by a factor of two or three can be
achieved at similar temporal and spatial resolution in
comparison with conventional segmented k-space
gradient echo imaging. Other potential applications
could be the real-time assessment of intimal flap
movement in aortic dissection delineating the relationship to branch vessels and possible occlusion of
these during systolic or diastolic phase. In addition,
differentiation between true and false lumen or possible thrombus and slow-flowing blood can be done
easily (Fig. 2).
A vulnerable point in SSFP is patient-induced B0
inhomogeneities. A local shim volume should be
used to eliminate these effects locally. Although there
is increasing interest assessing global and regional
ventricular dimensions and cardiac functions, SSFP
has not been proved as the modality of choice for
Fig. 2. A 57-year-old man with unknown right-sided chest pain and raised central venous pressure. Exclusion of suspected
central, chest tumor. (A) Coronal view of venous three-dimensional MRA data set delineating lack of contrast filling in the vena
cava superior. (B) Sagittal view of cine-mode TrueFISP sequence allowing differentiation between completely obstructed vena
cava superior and partially thrombosed right atrium and slow-flowing blood. (C) Coronal maximum intensity projection of the
thoracic vessels in venous phase missing the vena cava superior and delineating collateral flow in azygos and cervicothoracic
veins. (D) TrueFISP four-chamber view showing almost complete clotting of right atrium.
34
MRA because, until now, cine MR imaging has relied

exclusively on single-slice techniques requiring multiple breathhold [44 46]. For instance, coverage of
the entire aorta required scan times of more than
10 minutes and variations in inspiratory depth led to
discontinuous coverage of the data sets. Recently
developed multislice real-time SSFP cine sequences
capable of covering the entire thoracic or abdominal
aorta in the axial plane promises to resolve these
problems. For this technique, ECG triggering is
needed to synchronize the slices [47].
Another, more troublesome limitation of SSFP is
its enhancement of fat in addition to fluids. This
hinders the postprocessing of maximum intensity
projections. Because of the impact this has on image
evaluation, the usefulness of real-time cine SSFP in
vessel imaging remains doubtful.
Image analysis
The high contrast between Gd-containing luminal
(bright) and extraluminal (dark) spins and the true
three-dimensional nature of the acquired data sets
provides the basis for using a variety of postprocessing algorithms. Analysis should never be limited to
maximum intensity projections or surface-shaded
display. The three-dimensionality inherent to the
technique can only be exploited fully if the data are
viewed interactively on a workstation using multiplanar reformations. Endoluminal, virtual angioscopic
images can also be obtained, but their clinical relevance is very limited.
Clinical applications
Contrast-enhanced three-dimensional MRA has
proved to be a versatile noninvasive imaging modality for analysis of thoracic arteries and veins. In
the following sections various imaging indications
involving the thoracic aorta, the pulmonary arteries
and pulmonary veins, and the thoracic systemic veins
are discussed.
Thoracic aorta
Although CT remains the modality of choice in all
acute, life-threatening conditions involving the thoracic aorta, contrast-enhanced three-dimensional
MRA has emerged as the imaging modality of choice
for assessing the thoracic aorta in the more stable
patient. The technique overcomes respiratory and
cardiac motion artifacts, which impair noncontrast
MRA image quality to the point of rendering scans of

the thoracic aorta nondiagnostic. The underlying
three-dimensional GRE sequences permit data
acquisition of large fields-of-view in apnea and do
not require cardiac gating. Pulsatility artifacts, however, sometimes make it difficult to assess the supravalvular portion of the ascending thoracic aorta.
Aortic dissection
Beyond establishing the presence of dissection, it
is crucial to define the localization and extent of
disease. In fact, the location and relationship of the
intimal tear are critical to the choice of subsequent
therapeutic management. The Stanford classification
separates aortic dissections affecting the ascending
aorta (Stanford A) from those merely affecting the
descending aorta (Stanford B). Type A dissections
harbor the risk of myocardial infarction because of an
extension of the dissection into the coronary arteries
and of pericardial tamponade secondary to aortic
rupture into the pericardium. Type A dissections
require emergent surgery. Dissections arising distal
to the left subclavian (Stanford B) are less precarious
and are usually managed medically.
The efficacy of three-dimensional MRA in the
assessment and follow-up of aortic dissection is well
established. Three-dimensional MRA of the thoracic
aorta is fast and combines the advantages of arterial
contrast, similar to conventional catheter angiography,
with cross-sectional information. Using multiplanar
reformations, three-dimensional MRA provides a
comprehensive analysis in suspected aortic dissection:
the extent and relationship to branch vessels can be
depicted fully, and the true lumen can be separated
from the false lumen (Fig. 3).
The three-dimensional data set should be obtained
in the orientation optimal for visualization of the
target structure. The diagnostic value of three-dimensional contrast MRA is limited to the assessment of
the aortic lumen. Because aortic dissection can occur
without an intimal flap (intramural hematoma) or
inflammatory diseases (aortitis), an additional delayed
T1-weighted sequence, collected following contrast
administration, is recommended. For this purpose, the
three-dimensional contrast MRA sequence can be
repeated 2 minutes following the contrast administration or alternatively a T1-weighted GRE or spin echo
sequence can be acquired in the axial plane. Although
ECG gating is not necessary for the collection of a
three-dimensional data set, T1-weighted GRE or spin
echo sequences need to be used in conjunction with
ECG gating. On T1-weighted spin echo images, the
intramural hematoma is identified as concentric thickening of the aortic wall with increased intramural
35
Fig. 3. A 64-year-old man with acute thoracic pain. (A) Maximum intensity projection reformats of coronal three-dimensional
MRA data set (oblique sagittal plane) demonstrate aortic dissection distal to the origin of the left subclavian artery. (B) Axial
multiplanar reformation. (C) Sagittal multiplanar reformation.
signal intensity. Inflammatory processes also show

enhancement of the aortic wall and surrounding soft
tissues [10].
Additional functional information can be gathered with the use of cine imaging. Cine phase
contrast acquisitions permit measurement of flow
velocities and flow volumes and can easily differentiate between true and false lumen. Even aortic
valve involvement caused by dissection can be
evaluated. The severity of valvular regurgitation
can be determined with cine gradient echo MR
imaging, which allows measurement of the area of
the signal void corresponding to the abnormal flow
jet. Alternatively, this modality can be used to obtain
ventricular volumetric measurements and calculate
the regurgitant fraction or velocity-encoded cine.
MR imaging can be used to quantify regurgitant
blood flow [48].
Aortic aneurysm
Thoracic aortic aneurysms may be classified
according to their location, etiology, or shape. Natural history studies of thoracic aortic aneurysms
report a 1- and 5-year survival of 39% to 52%
and 13% to 19%, respectively [49]. Most mortality
in patients with thoracic aortic aneurysms is related
to aneurysm rupture. Thoracic aortic aneurysm is a
highly lethal condition warranting consideration of
elective, prophylactic surgical repair. The timing of
surgery is often a difficult clinical decision, however, particularly in asymptomatic patients or those
with comorbid conditions.
A recommendation of surgery represents a balance. Clinicians must weigh estimates of thoracic
aortic aneurysm natural history and rupture risk
against operative mortality and complication rate.
This risk is related to the site, etiology, size, and
expansion rate of the aneurysm. MR imaging of an
aortic aneurysm can demonstrate the site of aneurysm, its length, morphology, and relationship of the
aneurysm to branch vessels and the presence of mural
thrombus or a penetrating ulcer. All of these findings
can affect surgical decision making. Furthermore, the
aortic valve needs to be assessed for the presence of
valvular stenosis or insufficiency.
Complex underlying arterial morphology in
patients with thoracic aortic aneurysm may result in
an inadvertent exclusion of important portions of the
arterial anatomy from the three-dimensional imaging
volume. It is important to conduct the localizing process carefully using breathhold techniques
(Fig. 4). For imaging aortic aneurysms, which contain
slow flow, it is important to anticipate a long contrast
travel time to fill the entire aorta. As mentioned
previously, three-dimensional MRA contains little
information about the morphology of the aortic wall,
and should be complemented by T1-weighted postcontrast images in diagnosis of an aneurysm.
36
Fig. 4. Excessive aneurysm of the ascending aorta. (A) Sagittal maximum intensity projection projection demonstrates the
aneurysm originating immediately above the aortic valve and involving the supra-aortic vessels. (B) Axial half-Fourier singleshot turbo spin echo sequence delineating an extensive wall-adherent thrombus in the posterior part of descending aorta. No
dissection was observed.
Stenosis or occlusion of the great vessels may be

caused by atherosclerosis, dissection, and arteritis.
Other rare causes include fibromuscular dysplasia,
postradiation arteritis, and mediastinal inflammatory
or neoplastic disease. The patient commonly
presents with upper extremity ischemia or neurologic symptoms from the steal phenomena during
arm exercise. Coronal three-dimensional Gd-MRA
is the best sequence to assess origins of the innominate artery, common carotids, vertebral, and subclavian arteries.
Developmental abnormalities
MR imaging can safely assess less common
congenital cardiovascular malformations of the aorta,
including arch anomalies and aortic coarctation.
Coarctation may be described as preductal, juxtaductal, or postductal. The more common juxtaductal or
postductal types occur as a discrete focal narrowing

of the aortic isthmus distal to the origin of the left
subclavian artery and near the aortic end of the
ductus arteriosus (Fig. 5). Depending on the severity
of the coarctation and obliteration of ductus arteriosus, an abundance of collateral vessels may be seen.
MR imaging and MR imaging velocity mapping
have proved to visualize the anatomy and severity
of the coarctation accurately [50]. Gd-enhanced
three-dimensional MRA has the advantage for image
reconstructions in any desirable orientation providing
an accurate overview especially useful in tortuous
vessel structures. An apparent coarctation may be
recognized as a pseudocoarctation. The rapid acquisition time enables successful contrast-enhanced MRA
in children including neonates and infants [51].
Quantitative measurements can be obtained by using
velocity-encoded cine MRA [52].
Fig. 5. A 28-year-old man with coarctation of the aorta. Note the dilated proximal left subclavian artery. (A) Sagittal maximum
intensity projection projection. (B) Corresponding sagittal multiplanar reformation.
Central thoracic veins

Thrombosis of systemic chest veins is an important cause of morbidity in patients with malignancy,
hematologic disease, or long-term indwelling catheters. Prompt diagnosis and adequate therapy need to
be provided to restore patency of the veins.
Thrombo-occlusive disease of the chest veins has
been evaluated accurately by MR imaging using twodimensional time-of-flight methods [53,54]. These
techniques have the advantage of not requiring contrast material but are limited by long examination times
and potentially misleading artifacts [53,55]. Recent
reports have proposed the use of Gd-enhanced threedimensional MR venography techniques [56]. With
the advent of high-performance gradient systems, data
collection times have been reduced to acquire a threedimensional data set within a breathholding after the
intravenous injection of contrast agent.
In general, two approaches are pursued. Initially,
direct MR venography is performed using an injection of diluted contrast media to avoid T2 shortening
effects bolus. In comparison with two-dimensional
time-of-flight techniques, several recently published
studies show better image quality within shorter
imaging times when evaluating deep venous thrombosis [38]. Thornton et al [57] obtained breathholding
three-dimensional spoiled GRE during first pass and
in the delayed arteriovenous phase after manual
intravenous Gd bolus injection resulting in 100%
sensitivity and specificity. First-pass imaging, however, requires venous access site in the clinically
symptomatic limb and does not allow for a complete
evaluation of the chest veins, even with bilateral
infusion [58].
With an indirect approach, the veins are imaged
during the contrast equilibrium phase following the
injection of paramagnetic contrast into an antecubital
37
vein. To compensate for considerable dilution as the

contrast passes through the lungs, the arterial system,
and the capillary bed, indirect MR venography
requires large doses of contrast. The indirect
approach does not require cannulation of the vein in
the effected extremity. For better contrast-to-noise,
images obtained in the venous phase can be subtracted from those acquired in the arterial phase [59].
The timing of the acquisition relative to the contrast
administration is crucial. Kroenke et al [58] extrapolated the mean time of maximum contrast enhancement of the thoracic veins and chose a time delay of
15 seconds between the acquisition of the arterial and
venous phase. The introduction of dynamic Gdenhanced three-dimensional MR venography, based
on the use of very short TR (1.6 millisecond) and TE
(0.6 millisecond) [60] permits the acquisition of six
three-dimensional data sets in under 24 seconds. This
method obviates the need for contrast bolus timing
and has been shown to be reliable for the display of
central veins. This type of dynamic imaging also
allows for the assessment of thoracic veins in different arm positions (Fig. 6).
Another MR venography approach using a highresolution TrueFISP imaging sequence promises
visualization of thrombus with high contrast relative
to the surrounding blood pool without requiring the
administration of contrast agents. Limited investigations of patients with deep vein thrombosis have
been promising [61].
Pulmonary arteries
MR imaging of pulmonary vasculature has been
challenging for a variety of reasons. Particularly,
respiratory motion and poor contrast between flowing
blood and emboli have contributed toward poor
results [62]. Faster gradient hardware combined with
the dynamic administration of paramagnetic contrast
Fig. 6. A 29-year-old male conductor with left-sided arm weakness and known occlusion of subclavian and internal jugular vein
of the right side. Left-sided functional subclavian vein stenosis at the transition to brachiocephalic vein was suspected and
proofed using dynamic three-dimensional MRA. (A) Coronal maximum intensity projection (MIP), early phase. (B) Coronal
MIP, late phase.
38
Fig. 7. A 47-year-old woman with left pleuritic chest pain.

Coronal maximum intensity projection of a time-resolved
contrast-enhanced three-dimensional MRA data set of the
pulmonary arteries shows an embolus in the distal left
artery (arrow).
permits display of the pulmonary vasculature. Two

MR imaging-based approaches to assess the pulmonary arteries have been suggested: a sagittal acquisition strategy using two separate examinations
requiring twice the contrast volume versus the
acquisition of a single coronal three-dimensional
volume set encompassing both lungs [63 66]. The
advantage of sagittal imaging lies in the reduced
imaging volume, which permits shortening of the

data acquisition time. The coronal acquisition
approach exploits a single large field of view that
encompasses both lungs and has the advantage of
visualizing both central and peripheral pulmonary
vasculature (Fig. 7). Reflecting ease of interpretation
most investigators today would use the coronal data
acquisition technique.
By reducing TR to less than 2 milliseconds and
TE to less than 1 millisecond, temporally resolved
three-dimensional MRA becomes possible. The
acquisition time of the entire pulmonary tree can be
reduced to less than 4 seconds using the latest
hardware and software for the collection of a single
coronal three-dimensional data set [33]. Even in
patients with respiratory distress and limited breathholding capabilities the data collection time is sufficiently short to permit artifact-free depiction of the
pulmonary arterial tree. Although spatial resolution
has remained limited in most implementations, contrast-enhanced three-dimensional MRA was shown to
be capable of detecting subsegmental emboli.
A major advantage of MRA over alternative diagnostic strategies for investigation of patients with
suspected thromboembolic disease relates to the fact
that MRA of the pulmonary vasculature can be complemented by MR venography of the pelvic and
femoral veins. Vascular anomalies including arteriovenous malformations, anomalous pulmonary veins
and pulmonary sequestration, patent ductus arteriosus,
or pulmonary atresia associated with congenital heart
disease can also be detected (Fig. 8) [67]. Pulmonary
hypertension documented by three-dimensional MRA
provides a better delineation of the central pulmonary
arteries than conventional angiography, where the
catheter may be advanced into the pulmonary arteries
beyond significant pathologies.
Fig. 8. A 49-year-old man with suspected pulmonary arteriovenous malformation on chest radiograph in the right lower lobe.
The three-dimensional MRA data set revealed the feeding vessels and sharp delineated small nodules. The diagnosis was OslerWeber-Rendu disease. (A) Coronal maximum intensity projection reconstruction. (B) Shaded surface reconstruction
anteroposterior view of the arteriovenous malformation.
Illustration of the pulmonary veins is also provided

by contrast-enhanced three-dimensional MRA.
Venous structures are seen if the central k-space lines
are obtained during the venous or equilibrium phase of
the contrast bolus. The timing of the central k-space
views relative to contrast administration is crucial. The
need for timing can be overcome by collecting multiple three-dimensional data sets in various vascular
phases using sequential three-dimensional imaging
with ultrafast three-dimensional sequences. For the
optimal display of the pulmonary veins, a data set
containing both pulmonary veins and arteries is subtracted from a solely arterial data set [68]. Better
separation of arteries and veins can be achieved using
higher infusion rates and lower contrast media doses in
multiphase MR imaging protocols. Visualization of
main pulmonary veins, segmental veins, and subsegmental veins up to the fourth order can be achieved.
Summary
Using the described strategies all relevant disease
processes of the thoracic vessels can be fully
depicted using contrast-enhanced three-dimensional
MRA. The aorta and the major neck and arm vessels
are well visualized. Vascular pathologies, such as
aneurysms, dissections, and occlusions, are readily
recognized. With the implementation of high-performance gradients, three-dimensional MRA of the
pulmonary vasculature has become possible even in
dyspneic patients. Congenital lesions, such as coarctations, are particularly well suited for analysis with
these techniques.
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
References
[17]
[1] Prince MR, Yucel EK, Kaufman JA, et al. Dynamic
gadolinium-enhanced three-dimensional abdominal
MR arteriography. J Magn Reson Imaging 1993;
3:877 81.
[2] Leung DA, McKinnon GC, Davis CP, et al. Breathhold, contrast-enhanced, three-dimensional MR angiography. Radiology 1996;200:569 71.
[3] Prince MR, Narasimham DL, Stanley JC, et al. Breathhold gadolinium-enhanced MR angiography of the abdominal aorta and its major branches. Radiology
1995;197:785 92.
[4] Hartnell GG, Finn JP, Zenni M, et al. MR imaging of
the thoracic aorta: comparison of spin-echo, angiographic, and breathhold techniques. Radiology 1994;
191:697 704.
[5] Krinsky GA, Rofsky NM, DeCorato DR, et al. Thoracic aorta: comparison of gadolinium-enhanced three-
[18]
[19]
[20]
[21]
39
dimensional MR angiography with conventional MR

imaging. Radiology 1997;202:183 93.
Leung DA, Debatin JF. Three-dimensional contrastenhanced magnetic resonance angiography of the
thoracic vasculature. Eur Radiol 1997;7:981 9.
Goyen M, Ruehm SG, Debatin JF. MR-angiography:
the role of contrast agents. Eur J Radiol 2000;34:
247 56.
Shellock FG, Kanal E. Safety of magnetic resonance
imaging contrast agents. J Magn Reson Imaging
1999;10:477 84.
Hany TF, McKinnon GC, Leung DA, et al. Optimization of contrast timing for breathhold three-dimensional MR angiography. J Magn Reson Imaging 1997;
7:551 6.
Prince MR. Gadolinium-enhanced MR aortography.
Radiology 1994;191:155 64.
Prince MR, Chenevert TL, Foo TK, et al. Contrastenhanced abdominal MR angiography: optimization
of imaging delay time by automating the detection of
contrast material arrival in the aorta. Radiology 1997;
203:109 14.
Prince MR. Contrast-enhanced MR angiography:
theory and optimization. Magn Reson Imaging Clin
N Am 1998;6:257 67.
Korosec FR, Frayne R, Grist TM, Mistretta CA. Timeresolved contrast-enhanced 3D MR angiography.
Goyen M, Quick HH, Debatin JF, et al. Whole-body
three-dimensional MR angiography with a rolling table
platform: initial clinical experience. Radiology 2002;
22:270 7.
Meaney JF, Ridgway JP, Chakraverty S, et al. Stepping-table gadolinium-enhanced digital subtraction
MR angiography of the aorta and lower extremity arteries: preliminary experience. Radiology 1999; 211:
59 67.
Hany TF, Schmidt M, Hilfiker PR, et al. Optimization
of contrast dosage for gadolinium-enhanced 3D MRA
of the pulmonary and renal arteries. Magn Reson Imaging 1998;16:901 6.
Earls JP, Bluemke DA. New MR imaging contrast
agents. Magn Reson Imaging Clin N Am 1999;7:
255 73.
Knopp MV, von Tengg-Kobligk H, Floemer F, Schoenberg SO. Contrast agents for MRA: future directions.
J Magn Reson Imaging 1999;10:314 6.
Lahti KM, Lauffer RB, Chan T, Weisskoff RM. Magnetic resonance angiography at 0.3 T using MS-325.
Magma 2001;12:88 91.
Taupitz M, Schnorr J, Wagner S, et al. Coronary MR
angiography: experimental results with a monomerstabilized blood pool contrast medium. Radiology
2002;222:120 6.
Goyen M, Lauenstein TC, Herborn CU, et al. 0.5 mol/L
Gd chelate (Magnevist) versus 1.0 mol/L Gd chelate
(Gadovist): dose- independent effect on image quality
of pelvic three-dimensional MR-angiography. J Magn
Reson Imaging 2001;14:602 7.
40
[22] Li D, Zheng J, Bae KT, et al. Contrast-enhanced magnetic resonance imaging of the coronary arteries: a review. Invest Radiol 1998;33:578 86.
[23] Zheng J, Li D, Cavagna FM, et al. Contrast-enhanced
coronary MR angiography: relationship between coronary artery delineation and blood T1. J Magn Reson
Imaging 2001;14:348 54.
[24] Carroll TJ, Korosec FR, Swan JS, et al. The effect
of injection rate on time-resolved contrast-enhanced
peripheral MRA. J Magn Reson Imaging 2001;14:
401 10.
[25] Svensson J, Petersson JS, Stahlberg F, et al. Image
artifacts due to a time-varying contrast medium concentration in 3D contrast-enhanced MRA. J Magn Reson Imaging 1999;10:919 28.
[26] Lee VS, Martin DJ, Krinsky GA, Rofsky NM. Gadolinium-enhanced MR angiography: artifacts and pitfalls. AJR Am J Roentgenol 2000;175:197 205.
[27] Prince MR, Narasimham DL, Jacoby WT, et al. Threedimensional gadolinium-enhanced MR angiography of
the thoracic aorta. AJR Am J Roentgenol 1996;
166:1387 97.
[28] Earls JP, Rofsky NM, DeCorato DR, et al. Breathhold single-dose gadolinium-enhanced three-dimensional MR aortography: usefulness of a timing examination and MR power injector. Radiology 1996;
201:705 10.
[29] Kreitner KF, Kunz RP, Kalden P, et al. Contrast-enhanced three-dimensional MR angiography of the
thoracic aorta: experiences after 118 examinations with
a standard dose contrast administration and different
injection protocols. Eur Radiol 2001;11:1355 63.
[30] Foo TK, Saranathan M, Prince MR, Chenevert TL.
Automated detection of bolus arrival and initiation of
data acquisition in fast, three-dimensional, gadolinium-enhanced MR angiography. Radiology 1997;
203:275 80.
[31] Fellner FA, Fellner C, Wutke R, et al. Fluoroscopically
triggered contrast-enhanced 3D MR DSA and 3D timeof-flight turbo MRA of the carotid arteries: first clinical
experiences in correlation with ultrasound, x-ray angiography, and endarterectomy findings. Magn Reson
Imaging 2000;18:575 85.
[32] Wilman AH, Riederer SJ, King BF, et al. Fluoroscopically triggered contrast-enhanced three-dimensional
MR angiography with elliptical centric view order: application to the renal arteries. Radiology 1997;205:
137 46.
[33] Goyen M, Laub G, Ladd ME, et al. Dynamic 3D MR
angiography of the pulmonary arteries in under four
seconds. J Magn Reson Imaging 2001;13:372 7.
[34] Hennig J, Scheffler K, Laubenberger J, Strecker R.
Time-resolved projection angiography after bolus injection of contrast agent. Magn Reson Med 1997;
37:341 5.
[35] Sodickson DK, McKenzie CA, Ohliger MA, et al. Recent advances in image reconstruction, coil sensitivity
calibration, and coil array design for SMASH and generalized parallel MRI. MAGMA 2002;13:158 63.
[36] Frayne R, Grist TM, Swan JS, et al. 3D MR DSA:

effects of injection protocol and image masking.
[37] Neimatallah MA, Ho VB, Dong Q, et al. Gadoliniumenhanced 3D magnetic resonance angiography of the
thoracic vessels. J Magn Reson Imaging 1999;10:
758 70.
[38] Ruehm SG, Zimny K, Debatin JF. Direct contrastenhanced 3D MR venography. Eur Radiol 2001;11:
102 12.
[39] Sodickson DK, McKenzie CA, Li W, et al. Contrastenhanced 3D MR angiography with simultaneous
acquisition of spatial harmonics: a pilot study. Radiology 2000;217:284 9.
[40] Weiger M, Pruessmann KP, Leussler C, et al. Specific
coil design for SENSE: a six-element cardiac array.
Magn Reson Med 2001;45:495 504.
[41] Naganawa S, Koshikawa T, Fukatsu H, et al. Contrast-enhanced MR angiography of the carotid artery
using 3D time-resolved imaging of contrast kinetics:
comparison with real-time fluoroscopic triggered
3D-elliptical centric view ordering. Radiat Med
2001;19:185 92.
[42] Haacke EM, Tkach JA. Fast MR imaging: techniques
and clinical applications. AJR Am J Roentgenol
1990;155:951 64.
[43] Barkhausen J, Ruehm SG, Goyen M, et al. MR evaluation of ventricular function: true fast imaging with
steady-state precession versus fast low-angle shot cine
MR imaging: feasibility study. Radiology 2001;219:
264 9.
[44] Plein S, Bloomer TN, Ridgway JP, et al. Steady-state
free precession magnetic resonance imaging of the
heart: comparison with segmented k-space gradientecho imaging. J Magn Reson Imaging 2001;14:230 6.
[45] Thiele H, Nagel E, Paetsch I, et al. Functional cardiac
MR imaging with steady-state free precession (SSFP)
significantly improves endocardial border delineation
without contrast agents. J Magn Reson Imaging
2001;14:362 7.
[46] Bloomer TN, Plein S, Radjenovic A, et al. Cine MRI
using steady state free precession in the radial long axis
orientation is a fast accurate method for obtaining volumetric data of the left ventricle. J Magn Reson Imaging
2001;14:685 92.
[47] Barkhausen J, Goyen M, Ruhm SG, et al. Assessment
of ventricular function with single breathhold real-time
steady-state free precession cine MR imaging. AJR
Am J Roentgenol 2002;178:731 5.
[48] Didier D, Ratib O, Lerch R, Friedli B. Detection
and quantification of valvular heart disease with dynamic cardiac MR imaging. Radiographics 2000;20:
1279 301.
[49] Pitt MP, Bonser RS. The natural history of thoracic
aortic aneurysm disease: an overview. J Card Surg
1997;12(suppl 2):270 8.
[50] Holmqvist C, Stahlberg F, Hanseus K, et al. Collateral
flow in coarctation of the aorta with magnetic resonance velocity mapping: correlation to morphological
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
imaging of collateral vessels. J Magn Reson Imaging

2002;15:39 46.
Holmqvist C, Larsson EM, Stahlberg F, Laurin S. Contrast-enhanced thoracic 3D-MR angiography in infants
and children. Acta Radiol 2001;42:50 8.
Bogaert J, Kuzo R, Dymarkowski S, et al. Followup of patients with previous treatment for coarctation
of the thoracic aorta: comparison between contrastenhanced MR angiography and fast spin-echo MR
imaging. Eur Radiol 2000;10:1847 54.
Finn JP, Zisk JH, Edelman RR, et al. Central venous
occlusion: MR angiography. Radiology 1993;187:
245 51.
Hartnell GG, Hughes LA, Finn JP, Longmaid III HE.
Magnetic resonance angiography of the central chest
veins: a new gold standard? Chest 1995;107:1053 7.
Li W, David V, Kaplan R, Edelman RR. Three-dimensional low dose gadolinium-enhanced peripheral MR
venography. J Magn Reson Imaging 1998;8:630 3.
Lebowitz JA, Rofsky NM, Krinsky GA, Weinreb JC.
Gadolinium-enhanced body MR venography with subtraction technique. AJR Am J Roentgenol 1997;169:
755 8.
Thornton MJ, Ryan R, Varghese JC, et al. A threedimensional gadolinium-enhanced MR venography
technique for imaging central veins. AJR Am J Roentgenol 1999;173:999 1003.
Kroencke TJ, Taupitz M, Arnold R, et al. Three-dimensional gadolinium-enhanced magnetic resonance venography in suspected thrombo-occlusive disease of
the central chest veins. Chest 2001;120:1570 6.
Shinde TS, Lee VS, Rofsky NM, et al. Three-dimensional gadolinium-enhanced MR venographic evaluation of patency of central veins in the thorax: initial
experience. Radiology 1999;213:555 60.
41
[60] Goyen M, Barkhausen J, Kuehl H, et al. [Contrastenhanced 3D MR venography of central thoracic veins:
preliminary experience]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2001;173:356 61.
[61] Spuentrup E, Buecker A, Stuber M, Gunther RW. MRvenography using high resolution True-FISP. Rofo
Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr
2001;173:686 90.
[62] Kauczor HU, Gamroth AH, Tuengerthal SJ, et al. [MR
angiography: its use in pulmonary and mediastinal
space-occupying lesions]. Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1992;157:15 20.
[63] Gupta A, Frazer CK, Ferguson JM, et al. Acute pulmonary embolism: diagnosis with MR angiography.
Radiology 1999;210:353 9.
[64] Kauczor HU, Heussel CP, Thelen M. Update on diagnostic strategies of pulmonary embolism. Eur Radiol
1999;9:262 75.
[65] Meaney JF, Johansson LO, Ahlstrom H, Prince MR.
Pulmonary magnetic resonance angiography. J Magn
[66] Wielopolski PA, Hicks SG, de Bruin HG, Oudkerk M.
Breath-hold three-dimensional lung perfusion imaging
and pulmonary angiography after contrast administration. In: Oudkerk M, Edelman RR, editors. Highpower gradient MR-imaging: advances in MRI. Oxford: Blackwell Science; 1997. p. 71.
[67] Goyen M, Ruehm SG, Jagenburg A, et al. Pulmonary
arteriovenous malformation: characterization with
time-resolved ultra-fast 3D MR angiography. J Magn
[68] Schoenberg SO, Knopp MV, Grau A, et al. [Ultrafast
MRI phlebography of the lungs]. Radiologe 1998;
38:597 605.
Breast MR imaging
Mitchell D. Schnall, MD, PhD
Department of Radiology, Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104, USA
Breast cancer is a significant health care problem

in the United States. More than 180,000 American
women are diagnosed with breast cancer each year,
and approximately 50,000 of these women will die of
their disease [1]. Breast cancer is the second leading
cause of cancer death among women. Imaging plays
a crucial role in all aspects of breast cancer care. This
includes early detection through screening, diagnosis
and associated image-guided biopsy, treatment planning, and follow-up. The limitations of current x-ray
mammography have led to extensive efforts over the
past 15 years to develop complimentary imaging
techniques to improve breast imaging performance,
particularly in the radiographically dense breast. The
most accepted adjunct modality is breast sonography,
which is now widely used in the diagnostic evaluation of women with abnormal screening mammography or clinical exams. Other techniques that have
been proposed that are less widely used include
scintography with Tc99 Sestamibi and 18[F]-fluorodeoxyglucose positron emission tomography.
Early in its history, MR imaging was proposed as
a technique to assist in the detection and diagnosis of
breast cancer. Early reports of breast MR imaging
confirmed that high-quality MR images of the breast
could be obtained with local surface coils [2,3]. In
the absence of exogenously injected contrast agents,
however, it was difficult to detect breast cancer. In
1989, Kaiser and Zeitler [4] and Heywang et al [5]
independently published on the application of MR
imaging contrast agents to detect breast cancer. Their
findingsthat the use of intravenous gadolinium
chelates allow MR imaging to detect and possibly
E-mail address: schnall@rad.upenn.edu
diagnose breast cancerwere extremely exciting and

led to significant follow-up work.
Using a higher resolution, three-dimensional (3D)
technique, Harms et al [6] demonstrated the power
of MR imaging to detect mammographically and
clinically occult breast cancer. They performed careful correlations between examinations performed on
women prior to mastectomy and the resultant pathology, and showed that high-resolution MR imaging,
performed with a technique pioneered in his laboratory rotating delivery of excitation off resonance
(RODEO), was able to detect occult multifocal
cancer in up to 40% of women.
Despite the extremely high sensitivity for breast
cancer MR imaging demonstrated in these early
studies, it was clear that contrast enhancement alone
was not specific for breast cancer. This led to intense
efforts to identify distinguishing characteristics
between benign enhancing lesions and malignant
enhancing lesions. The use of information obtained
from the architecture of the enhancing lesion and
qualitative and quantitative interpretations of the
pharmicokinetics of enhancement have been studied
extensively for this purpose. Although many different
pharmacokinetic imaging approaches have been
described, the most commonly adopted method is
the qualitative approach to the time signal intensity
curve of gadolinium, which was popularized by Kuhl
et al [7] and Kaiser et al [4]. Work by Orel et al [8]
and Nunes et al [9] clearly demonstrated the importance of lesion architecture for distinguishing between benign and malignant lesions. Today most
practitioners agree that a combination of gadolinium
pharmokinetics and lesion architecture is important
for proper interpretation of breast MR images.
Recently, breast MR imaging has become more
widely used as a supplemental imaging modality in
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 8 - 4
44
M.D. Schnall / Radiol Clin N Am 41 (2003) 4350
the breast. The most common clinical indications for

which breast MR imaging is being performed include difficult diagnostic evaluations and the evaluation of extent of disease within a breast affected
with breast cancer. This article discusses breast MR
imaging techniques, interpretation strategies, and
strategies for clinical implementation.
Breast MR imaging techniques

Breast MR imaging should be performed using a
dedicated breast surface coil. There are commercially
available unilateral and bilateral breast surface coils.
Most currently available surface coils also allow for
some access to the breast for image-guided procedures. Similar to mammography of the breast, MR
imaging requires breast compression for optimal
imaging. The compression also provides stabilization
to prevent motion and reduces the breast size in one
dimension reducing the requirement for coverage in
the medial to lateral dimension.
Almost all reports of breast MR imaging in the
literature have been performed at 1.0 or 1.5 T.
Although there have been scattered reports of breast
MR imaging being performed successfully at 0.5 T
[10], there is no scientific literature regarding the
performance of breast MR imaging at field strengths
less than 0.5 T. Due to the technical demands of
breast MR imaging and the physical properties of
gadolinium, optimal imaging performance is attained
using increased MR field strengths. There is no
clinical data, however, that provides a guide to the
relative clinical performance of breast MR at different
field strengths. Despite this, I would exercise great
caution at performing contrast-enhanced breast MR
imaging at field strengths less than 0.5 T.
There is no specific patient preparation for breast
MR imaging. Patients should undergo standard MR
imaging screening procedures. Even though metal
objects may be far from the imaging volume, they
may affect the ability to fat suppress; thus, all metallic
objects should be removed from the patients body,
including zippers and clasps. It is best to place an IV
for gadolinium injection prior to the patient entering
the magnet suite. An IV line should be established so
that the injection of gadolinium can be performed
without moving the patient. Gadolinium injection
may be performed by a mechanical injector or by hand.
After obtaining pilot images, I routinely perform
T2-weighted images. Although in the majority of
cases, the T2-weighted images provide no added
diagnostic value, they do provide information regarding the characterization of cystic lesions and fibro-
adenomas in some cases. Given the small time

commitment necessary to obtain these images, I
would advise routinely obtaining T2-weighted
images. I obtain T2-weighted images as fat-suppressed, fast spin echo images, using a 256 192
matrix over an 18-cm field of view. I use a 3-mm
slice thickness. The repitition time (TR)/echo time
(TE) of the sequence that I use is approximately
4 seconds per 90 milliseconds.
The most critical part of the breast MR imaging
acquisition is the contrast-enhanced T1-weighted
acquisition. This consists of contrast-enhanced images
performed prior to and after the administration of
gadolinium. The precontrast and postcontrast acquisition should adhere to certain principles. There should
be sufficient T1 weighting to show differences in
contrast enhancement, even at high contrast concentrations. Three-dimensional acquisitions with very
short TRs will easily meet this requirement. The same
pulse sequence should be used prior to and after the
administration of contrast. There should be no change
in system operating parameters between precontrast
and postcontrast images, so that direct subtractions and
quantitative comparisons may be performed. A minimum of two sequential postcontrast acquisitions
should be performed. The postcontrast imaging should
be initiated dynamically during or after the infusion of
contrast agent. Although there is no clear consensus on
the required time resolution of postcontrast images, I
recommend 3 minutes as an upper limit for the time
resolution of the dynamic gadolinium study.
Fat suppression is a valuable aid to improve the
conspicuity of contrast enhancement. Most modern
scanners can perform fat suppression over a volume
that encompasses both breasts. Chemical shift
selective inversion performed intermittently during
a 3D echo sequence provides a time-efficient
method for performing fat suppression. There is
relatively little time penalty in obtaining fat suppression in this manner.
During image interpretation, subtraction can further enhance the conspicuity of enhancing lesions.
This is considered mandatory if no fat suppression is
employed. Even if fat suppression is employed,
subtraction can be valuable to offset inhomogeneities
in fat suppression or to add further conspicuity to
enhancing lesions.
Quantitative image enhancement and the effect

of gadolinium
As discussed in the introduction, there have
been many efforts to establish quantitative criteria
based on enhancement profiles to differentiate

benign from malignant enhancing lesions. The
complex relationship between gadolinium concentration, pulse sequence, and signal intensity makes
it extremely difficult for quantitative techniques to
be developed that are generalizable across different
platforms. To appreciate this issue, it is important
to first understand the mechanism by which gadolinium causes enhancement of signal intensity on
MR imaging.
The mechanism of action of gadolinium is fundamentally different than what radiologists are
accustomed to with iodinated contrast agents. In
MR imaging, the gadolinium itself is never actually
visualized. Enhancement of the MR signal intensity
is derived from the water within the tissue due to
the effect of gadolinium interaction with water in
tissue. This interaction leads to a shortening of
the T1 time and makes water more visible on a
T1-weighted image. The resultant increased signal
intensity is a complex function of the concentration
of the gadolinium delivered to the tissue, the inherent T1 time of the tissue prior to enhancement, and
the pulse sequence itself. It is valuable to qualitatively understand how each of these factors affects
the enhancement.
As stated above, the mechanism of action of
gadolinium is to change the T1 relaxation time of
the water within the tissue of the breast. Typical
breast tissue and breast lesions have T1 values that
vary from 700 to 1000 milliseconds. When the
gadolinium is injected, the tissue enhances and the
T1 value of that tissue (at 1.5 T) is reduced to
approximately 200 milliseconds. Therefore, the signal intensity of breast tissue after contrast enhancement is unrelated to the initial T1 value of the
gadolinium concentration. If a signal intensity ratio
were to be calculated (S postcontrast/S precontrast), different values would be obtained for different tissues
with different precontrast T1 values, despite the
same gadolinium concentration delivered to each
tissue. This is because the denominator would be
different based on the initial T1. Therefore, it is
clear that the precontrast T1 value will influence any
signal enhancement ratio that is calculated during a
contrast-enhanced breast MR imaging examination.
The pulse sequence also has a large influence on
the signal enhancement ratio. All lectures on basic
MR imaging physics include a discussion of how
short TR times result in T1 weighting. Similarly, for
gradient echo images, larger flip angles increase T1
weighting. The ratio of these signals for tissues with
different T1 values will vary according to the pulse
sequence used. Therefore, a signal enhancement
45
ratio obtained from one particular pulse sequence

will not be directly generalizable to other pulse
sequences. Therefore, signal enhancement ratio criteria for diagnosing breast lesions under MR
imaging cannot be generalized across institutions
and platforms.
A more generalizable solution is to directly
calculate the gadolinium concentration delivered to
the tumor and use quantitative criteria that rely on
the gadolinium concentration itself. This allows for
quantitation of each case, independent of the pulse
sequence used. This also corrects for the influence
of the initial lesion T1 on the signal enhancement ratio.
There are discussions in the literature that use
basic physiologic parameters derived from the gadolinium concentration or pharmacokinetic curve to
classify breast MR lesions [11,12]. Most of these
calculations assume a linear relationship between
relaxation time and gadolinium concentration. This
would be the case if there were a single compartment within the tissue over which the gadolinium
mixed evenly; however, the distribution of gadolinium within a tissue is more complicated. Gadolinium extravasates from blood vessels into the
extracellular space, but does not enter cells themselves. Therefore, the water that is inside cells does
not have the same access to gadolinium as the water
that is outside cells. Thus, tissue is characterized by
two separate relaxation timesan intracellular
relaxation time and an extracellular relaxation time.
If there is rapid exchange of the water across the
cell membrane (rapid being defined as fast compared with the difference in relaxation rates between
the compartments), a linear relationship between
gadolinium concentration and average T1 can be
assumed, causing no error in the gadolinium concentration calculation. Landis et al [13] recently
showed, however, that this is not the case at peak
tissue gadolinium concentrations after a dynamic
injection of gadolinium. The high concentration of
gadolinium in the extracellular space of tumors
makes the assumption of rapid exchange invalid.
This creates a complicated relationship between the
signal intensity ratio and the gadolinium concentration. Correction for this effect is not trivial, but
is possible. The point to be understood is that developing quantitative criteria based on signal enhancement characteristics of lesions after dynamic
contrast injection that are accurate and reproducible
across multiple platforms is extremely difficult. Any
recommendation in the literature to apply a quantitative criterion should be reviewed in light of
these difficulties.
46
Image interpretation
Interpretation of breast MR images is best performed on a computer workstation where reformatting, image processing, image subtraction, and image
quantitation is available. Film presentations are difficult, due to the number of images that comprise a
breast MR imaging examination. This makes it difficult for patients to have images available for second
opinions at other institutions. It is suggested that at
least one early phase postcontrast image set that
includes the entire breast be available on film at the
patients request. In addition, selected images including reformatting and subtraction through any
significant findings should be available. Associated
enhancement curves should also be available on film
or paper copy if they are felt to be relevant. This is the
minimum information needed by a colleague in
another institution to assist in the patients care.
The major task in interpreting breast MR imaging
is establishing the likelihood of malignancy for any
enhancement observed. There are two major classes
of image features that are available to the radiologist
to assist in this interpretation. These are features

based on the architecture of the lesion enhancement
and features from the times/signal intensity curve of
the enhancement.
A structured approach to the use of architectural
features in breast MR imaging has been described by
Nunes et al [9]. This approach has been refined and
some of its aspects are being integrated into a breast
MR imaging lexicon being developed under the
auspices of the American College of Radiology
[14]. These approaches would suggest that the initial
approach to enhancement of breast is to make a
determination of whether the enhancement represents
a focal mass or not a focal mass. An example of focal
and nonfocal mass enhancement is shown in Fig. 1.
Data from Nunes et al [9] suggests that approximately
half of the focal masses identified in a diagnostic
population will be cancer. In order to define which of
these masses are malignant and which are benign,
other architectural features can be utilized. An
important feature is the shape and border of the
lesion (Fig. 2). The more irregular or spiculated the
lesion margin is, the more likely it is to be cancer.
Fig. 1. (A) Example of stippled regional enhancement from benign hyperplasia. (B) Lobulated focal mass, representing
a fibroadenoma.
47
Fig. 2. (A) Irregular-bordered enhancing mass secondary to invasive cancer. (B) Smooth-bordered enhancing mass, representing
a fibroadenoma.
Lesion borders that are smooth or demonstrate gentle

lobulations are more likely to be benign. This distinction is still not 100% accurate. There are lobulated-bordered cancers; for example, colloid, tubular,
and medullary cancers can have well-defined lobulated borders. In addition, benign lesions such as
radial scars can demonstrate spiculation. Other features can be valuable in further distinguishing
between benign and malignant enhancing focal
masses. Rim enhancement of the lesion is highly
suggestive of malignancy (Fig. 3). Note that this is
only the case for solid lesions. Cystic lesions in the
breast will typically have an enhancing rim. The
smooth enhancing rim of a cystic lesion does not
suggest any evidence of malignancy. It is noted that
enhancing rims in solid lesions can occur not only at
the periphery of the lesion, but occasionally can be
seen entering the internal portion of a lesion.
There are features that are most suggestive of
benignity. Fibroadenomas tend to grow in several
adjacent lobulas. They respect the demarcation of
these lobulas and develop a fibrous septum between
the various lobules. The septum can be seen as a lowsignal nonenhancing internal septation within a lesion.
This finding is seen on postcontrast images. In addition, fibroadenomas may be edematous and extremely
bright on T2-weighted images, particularly in young
women. Thus, the fibrous septum can also be seen on
T2-weighted images. The finding of septations within
a lobulated or smooth-bordered lesion should be
considered as good evidence of benignity (Fig. 4).
Although fibroadenomas in young women may be
edematous and bright on T2-weighted images, they

hyalinize over time. Hyalinized fibroadenomas tend to
be low signal on T2-weighted images. Although
cancers can occasionally be low signal, this is usually
due to extensive desmoplastic reaction and is associated with a spiculated lesion. A smooth border or
lobulated lesion that is low signal on T2 will almost
certainly represent a hyalinized fibroadenoma. These
hyalinized fibroadenomas may enhance or may eventually lose their blood supply and stop enhancing.
Fig. 3. Rim-enhancing mass, representing an invasive

carcinoma.
48
Fig. 4. Lobulated focal mass with internal septations,

representing a fibroadenoma.
Because fibroadenomas develop multiple lobules,

different lobulations in the fibroadenomas may have
different characteristics. Some are edematous whereas
others are hyalinized. This may lead to heterogeneous
appearances on both postcontrast and T2-weighted
images. Applying the principles above to each lobule
will allow for a diagnosis to be established.
There are two important characteristics of nonfocal mass enhancement that need to be determined.
The two main important factors to be determined
are the distribution of the enhancement and the form
of the enhancement. If the distribution resembles the
distribution of breast ducts, it is more likely to be
cancer. Distributions can be described as ductal,
segmental, regional, or diffuse. Descriptions will also
refer to the size of the foci that make up the area
of enhancement. These descriptions will include
stippled, clumped, inhomogeneous, and confluent.
Therefore, using these characteristics, stippled enhancement in a regional or diffuse distribution most
likely will be benign (Fig. 5), and confluent or
clumped enhancement in a segmental distribution
most likely will be malignant. Lesions identified as
nonfocal mass enhancement most often will include a
significant fraction of in situ cancer; however, there
may be associated invasion.
In addition to the architectural information described above, there are features related to the time
course of enhancement that are predictive of cancer.
As described earlier, the most robust and reproducible features relate to the qualitative assessment
of the enhancement curve. The enhancement curve is
not measured as the average enhancement curve over

the lesion. Rather, the enhancement curve should be
sampled from multiple locations in the lesion, and
the most suspicious enhancement curve in the lesion
should be assigned to the lesion. Enhancement
curves can be divided into three major types: persistent, plateau, and washout. Persistent enhancing
curves demonstrate continued enhancement beyond
the first 2 minutes of acquisition. Plateau curves will
plateau and level off after 2 minutes of contrast
injection. Washout curves will reach a peak after
2 minutes of contrast injection and new signal intensity (Fig. 6). Washout is felt to be a feature
suspicious for cancer, plateau is felt to be indeterminate, and persistent enhancement is reported to be
a feature most consistent with benignity. These
features are not 100% accurate in these determinations, and reported accuracies have varied. This is
particularly true in nonfocal mass enhancement
in which persistent enhancement does not exclude
malignancy. Similarly, classic fibroadenomas can
Fig. 5. Clumped enhancement in a segmental or ductal

distribution, representing ductal carcinoma in situ.
Fig. 6. Patient who has had a primary resection for invasive

cancer demonstrating irregular enhancing mass adjacent to
the resection cavity, representing additional foci of cancer.
49
breast conservation therapy. There is a concern that

breast MR imaging will lead to more extensive
surgery for breast cancer without a significant effect
on the recurrence rate or long-term survival. In effect,
MR imaging may be guiding the excision of foci of
tumor that could be treated with radiation therapy.
This issue needs to be studied in a prospective, wellcontrolled trial. In addition, the evaluation of the
contralateral breast is another significant issue that
needs to be studied. Reports in the literature suggest a
5% to 10% detection rate of cancer in the contralateral breast, using breast MR imaging at the time of
presentation of the initial cancer. The determination
of contralateral breast cancer has an affect on management strategy and is best determined at the time of
presentation rather than in subsequent years. Only
small, single-institution studies have been performed
to date. A larger, multi-institutional trial is needed to
clearly establish the role of breast MR imaging in
evaluating the contralateral breast.
Breast cancer screening

demonstrate washout and this should not be used as
evidence of malignancy. My approach is to initially
apply an architectural analysis, and then use the timecourse kinetics to either confirm a diagnosis or assist
interpretation in borderline cases.
Breast MR imaging evaluation of the extent of

breast cancer within an affected breast
One of the major clinical indications for breast
MR imaging is to determine the extent of cancer
within an affected breast. This use of MR imaging
was pioneered by Harms et al [6], who demonstrated
additional foci of cancer that were mammographically occult in up to 40% of women undergoing
mastectomy for breast cancer. Although this population was biased toward more advanced lesions by
virtue of the fact that these women were undergoing
mastectomy, this work highlighted a role for MR
imaging in the detection of more extensive disease.
There are multiple reports in the literature that suggest that breast MR imaging will effect the management of between 10% and 20% of women with newly
diagnosed breast cancer [15,16].
There are still many unanswered questions regarding the use of breast MR imaging in the evaluation of
the extent of breast cancer. Although it is clear that
MR imaging can detect additional foci of disease, it
has never been determined that MR imaging has an
effect on the recurrence rate of patients undergoing
The expense and technical resources required by

breast MR imaging in its current form makes it difficult to think of this technique as a general screening
technique. There has been tremendous progress in
quantifying an individual womens risk for breast
cancer [17,18], however. In high-risk patients, the
enhanced sensitivity of MR imaging may offer an
important adjunct to their screening regiment. Several
small pilot studies [19 22] have concluded that breast
MR imaging has the potential to detect mammographically and clinically occult breast cancer in high-risk
women. The reported yields for the MR imaging
detection of otherwise occult breast cancer in highrisk women is in the 1% to 3% range. These results
obviously would be highly dependent on patient
population and length of time over which the screening intervention is employed. Studies that have
included sonography in addition to mammography
and MR imaging also have shown that MR imaging
had a similar high yield relative to US in detecting
breast cancer in this population [7,20]. Although
there are many unanswered questions, a consensus
is forming that MR imaging may indeed have a role
in screening high-risk women for breast cancer. The
performance of breast MR imaging in a screening
role (receiver operating characteristic curve) has not
been well established. In addition, the manner in
which high-risk patients are likely to benefit from
MR screening also has not been answered. Other
questions such as the timing between MR scans and
50
at what age MR screens should be initiated remain

unanswered. Perhaps the most critical unanswered
question is the overall health impact of MR imaging
screening in this population. A large-scale screening
study involving high-risk women needs to be undertaken to answer these important questions.
Summary
Contrast-enhanced breast MR imaging has made
significant progress since its introduction into the
radiological literature in 1989. The techniques and
technology continue to be refined, and understanding
of the interpretation strategies has improved dramatically. Clinical applications in difficult diagnostic
cases and the evaluation of the extent of breast cancer
are now being practiced in many centers worldwide.
There is great excitement over the potential for breast
MR imaging to address the problem of screening
high-risk women. Despite all of the progress made
over the past years, however, there is still a significant amount of work ahead before a clear understanding of how this technique will affect the health
care of women is obtained.
References
[1] National Institutes of Health consensus development
conference statement: treatment of early breast cancer.
Bethesda, MD, June 18 21, 1990.
[2] Dash N, Lupetin AR, Daffner RH, et al. Magnetic
resonance imaging in the diagnosis of breast disease.
AJR Am J Roentgenol 1986;146:119 25.
[3] Stelling CB, Wang PC, Lieber A, et al. Prototype coil
for magnetic resonance imaging of the female breast.
Radiology 1985;154:457 62.
[4] Kaiser WA, Zeitler E. MR imaging of the breast: fast
imaging sequences with and without Gd-DTPA. Radiology 1989;170:681 6.
[5] Heywang SH, Wolf A, Pruss E, et al. MR imaging of
the breast with Gd-DTPA: use and limitations. Radiology 1989;171:95 103.
[6] Harms SE, Flamig DP, Hesley KL, et al. MR imaging
of the breast with rotating delivery of excitation off
resonance: clinical experience with pathologic correlation. Radiology 1993;187:493 501.
[7] Kuhl CK, Mielcareck P, Klaschik S, et al. Dynamic
breast MR imaging: are signal intensity time course
data useful for differential diagnosis of enhancing lesions? Radiology 1999;211(1):101 10.
[8] Orel SG, Schnall MD, LiVolsi VA, Troupin RH.
Suspicious breast lesions: MR imaging with radiologic pathologic correlation. Radiology 1994;190:
485 94.
[9] Nunes LW, Schnall MD, Orel SG, et al. Breast MR

imaging: interpretation model. Radiology 1997;202:
833 41.
[10] Kuhl CK, Kreft BP, Hauswirth A, et al. [MR mammography at 0.5 tesla. I. Comparison of image quality and
sensitivity of MR mammography at 0.5 and 1.5 T.].
Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr
1995;162(5):381 9.
[11] Tofts PS, Berkowitz B, Schnall MD. Quantitative analysis of dynamic Gd-DTPA enhancement in breast tumors using a permeability model. Magn Reson Med
1995;33(4):564 8.
[12] Hulka CA, Smith BL, Sgroi DC, et al. Benign and
malignant breast lesions: differentiation with echoplanar MR imaging. Radiology 1995;197(1):33 8.
[13] Landis CS, Li X, Telang FW, et al. Determination of
the MRI contrast agent concentration time course in
vivo following bolus injection: effect of equilibrium
transcytolemmal water exchange. Magn Reson Med
2000;44(4):563 74.
[14] Ikeda KK, Hylton NM, Kinkel K, et al. Development,
standardization, and testing of a lexicon for reporting
contrast-enhanced breast magnetic resonance imaging studies. J Magn Reson Imaging 2001;13(6):
889 95.
[15] Tan JE, Orel SG, Schnall MD, et al. Role of magnetic
resonance imaging and magnetic resonance imaging
guided surgery in the evaluation of patients with earlystage breast cancer for breast conservation treatment.
Am J Clin Oncol 1999;22(4):414 8.
[16] Fischer U, Kopka L, Grabbe E. Breast carcinoma:
effect of preoperative contrast-enhanced MR imaging
on the therapeutic approach. Radiology 1999;213(3):
881 8.
[17] Claus EB, Risch N, Thompson WD. Genetic analysis
of breast cancer in the cancer and steriod hormone
study. Am J Hum Genet 1991;48:232 42.
[18] Gail MH, Brinton LA, Byar DP, et al. Projecting individualized probabilities of developing breast cancer for
white females who are examined annually. J Natl Cancer Inst 1989;81:1879 86.
[19] Stoutjesdijk MJ, Boetes C, Jager GJ, et al. Magnetic
resonance imaging and mammography in women with
a hereditary risk of breast cancer. J Natl Cancer Inst
2001;93(14):1095 102.
[20] Warner E, Plewes DB, Shumak RS, et al. Comparison
of breast magnetic resonance imaging, mammography,
and ultrasound for surveillance of women at high risk
for hereditary breast cancer. J Clin Oncol 2001;19(15):
3524 31.
[21] Kuhl KK, Schmutzler R, Leutner CC, et al. Breast MR
imaging screening in 192 women proved or suspected
to be carriers of a breast cancer susceptibility gene:
preliminary results. Radiology 2000;215:267 79.
[22] Lo LD, Rosen MA, Schnall MD, et al. Pilot study of
breast MR screening of a high-risk cohort [abstract
843]. In: Scientific program of the Radiological Society of North America 87th Scientific Assembly and
Annual Meeting. Chicago: 2001. p. 432.
MR imaging of liver tumors

Ihab R. Kamel, MD, PhD*, David A. Bluemke, MD, PhD
The Russell H. Morgan Department of Radiology and Radiological Sciences, Johns Hopkins Hospital,
The advantages of MR imaging in the investigation of liver disease are well documented. The lack of
ionizing radiation of MR imaging and the safety of
gadolinium chelates are important considerations.
Unlike iodinated contrast agents, gadolinium chelates
have negligible effects on renal function in patients
with renal failure and are much less likely to result in
major allergic reactions [1]. Moreover, recent developments in MR imaging, including fast scanning
techniques and new MR imaging contrast agents,
enable improved detection and characterization of
many liver lesions so that a definitive diagnosis can
be made noninvasively. Lesion characterization and
determination of the extent of disease often have
important prognostic and therapeutic implications.
Since the early 1990s, reports comparing current
MR imaging and CT approaches have shown that
MR imaging is more accurate than contrast-enhanced
CT for the detection and characterization of liver
lesions [2 8]. Earlier publications compared MR
imaging to dynamic contrast-enhanced CT, whereas
later studies compared MR imaging to CT arterial
portography or spiral CT. Comparisons to multidetector CT have not been published yet, but the
two modalities are probably nearly equivalent in the
evaluation of intrahepatic disease.
The aim of this article is to review the features
of liver masses using current state-of-the-art MR
imaging techniques, and to discuss the role of new
liver-specific contrast agents in the detection and
characterization of liver masses.
E-mail address: ikamel@jhmi.edu (I.R. Kamel).
Liver imaging protocol

The routine protocol for screening MR imaging
of the liver consists of axial T1-weighted and
T2-weighted sequences. For high field strength
units, T2-weighted sequences are more sensitive to
intrahepatic disease, whereas for low field units,
T1-weighted sequences are more sensitive. Depending on the clinical question, acquisitions in the coronal
or sagittal plane should be added; for example, if there
is a question of a subdiaphragmatic process, these
additional planes may be more helpful then the axial
plane. For the evaluation of vascular structures, postcontrast axial gradient echo (GRE) sequence is performed in different phases of contrast enhancement.
T1-weighted sequences
Breath-hold spoiled GRE (SGE) is widely used as a
T1-weighted sequence for evaluating the liver
[4,9,10]. Parameters are repetition time (TR) 150 to
200/msec, echo time (TE) minimum, optional fat
suppression, 256 192, 3/4 field of view, 8 2 mm,
breath hold. This sequence allows imaging of the
entire liver during a single breath hold. Other advantages include good signal-to-noise ratio, strong T1
weighting, and minimal magnetic susceptibility
effects [11]. Imaging parameters for SGE appear to
be generalized to provide good diagnostic quality on
any equipment that has the appropriate gradient system [12,13]. An important feature of the multisection
acquisition of SGE is that the central-phase encoding
steps, which determine the bulk signal in the image,
are acquired over 6 seconds for both the entire data set
and each individual section. Thus, the data acquisition
is sufficiently short in order for the entire data set to
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 7 - 2
52
I.R. Kamel, D.A. Bluemke / Radiol Clin N Am 41 (2003) 5165
isolate a distinct phase of contrast enhancement. Each

individual section is sufficiently long, however, to
compensate for slight variations in injection technique,
patient cardiac output, and peak lesion enhancement.
Three-dimensional (3D) GRE imaging is being
investigated as a T1-weighted sequence to provide
accurate 3D segmental and vascular anatomy for
surgical planning, including assessment prior to liver
resection and transplantation [14]. The image quality
of 3D GRE continues to undergo refinement and is
quickly approaching the quality of two-dimensional
(2D) SGE [15].
T2-weighted sequences
Breathing-averaged T2-weighted conventional or
echo train spin echo, with optional fat suppression, is
the most commonly used T2-weighted sequence.
Parameters include TR 4000/msec, TE 100/msec,
7 10 mm interleaved, echo train length 8 to 16,
256 256, 2 number of excitation (NEX), respiratory trigger, superior/inferior saturation bands, 32 kHz,
fat suppression. In patients whose breathing is regular, this sequence generates images of high diagnostic
quality; however, the image quality of breathingaveraged sequences is inconsistent for patients who
may not be able to breathe regularly and breath-hold
T2-weighted images are employed. Breath-hold
imaging represents an opportunity to remove respiratory motion. Fast spin echo (FSE) techniques are
compatible with breath-hold acquisitions by using
long echo train lengths of approximately 20 and
reduced matrix array size. By using a high bandwidth
acquisition (eg, 32 kHz), acquisition times are further
reduced. Using this method, approximately eight
axial sections can be acquired throughout the liver
in 20 to 30 seconds. Complete imaging of the liver
can be completed in about 1 minute. This technique
has been found to have better contrast to-noise ratio
(CNR) than conventional spin echo sequences [16].
FSE breath-hold sequences are currently being evaluated in comparison with FSE without breath holding. Early experience suggests that FSE image
without breath holding is superior to breath-hold
FSE acquisitions.
Fat suppression improves the quality of both conventional and FSE T2-weighted sequences [17 20].
There are two reasons for this. First, moving fat
during respiration causes phase artifacts that are
propagated through the entire image. By using fat
suppression, the signal of the moving anterior abdominal wall is suppressed, and thus it contributes to
respiratory artifacts. Second, suppression of the high
signal fat (especially with FSE or turbo spin echo)
allows for a greater dynamic range of signal intensities to be displayed in the images. This can cause
increased conspicuity of focal lesions in the liver.
Because of local field inhomogeneities, however, it is
necessary to perform manual tuning of the fat suppression pulse on most systems. In addition, fat
suppression away from the isocenter of the magnet
may not be effective.
Half-Fourier acquisition single-shot turbo spin
echo (HASTE) or single-shot FSE sequence provides
breath-hold T2-weighted imaging of the liver. The
technique is acquired just over half of the k-space
in single echo train. The symmetry of the k-space
allows for mathematical reconstruction of the image
[21 23]. HASTE is helpful in patients who are
unable to breath hold because it can be acquired in
less than 1 second. Effectively the TR is infinite (or
undefined), whereas the TE can be manipulated to
achieve different contrasts. HASTE has been shown
to be effective for characterizing cysts and hemangiomas with long T2 times. HASTE is not effective
for tumor detection due to poor contrast resolution for
lesions with moderate T2 times. It should not be used
as the primary sequence for liver tumor detection.
Contrast-enhanced sequences
Intravenous contrast agents have been used in
evaluating the liver since the late 1980s. The use of
contrast agents can potentially increase the sensitivity
and specificity of liver MR imaging in detection and
characterization of liver lesions.
The first category of contrast agents to be used in
clinical practice was extracellular gadolinium chelates. Two other classes of contrast agents, reticuloendothelial system (RES) specific and hepatocyte
selective, recently have been developed and employed
for liver studies. In addition, contrast agents that have
combined perfusional and hepatocyte selective functions recently have been introduced, and their utility is
discussed below.
Extracellular gadolinium chelates
There are two physiological principles that allow
targeting contrast enhancement: the dual blood supply
of the liver and the hemodynamics of hepatic tumors.
The liver is unique among abdominal organs in having
a dual blood supply, with the portal vein supplying
75% to 80% of flow and the hepatic artery supplying
the remaining 20% to 25%. Liver tumors, however,
receive nearly all their blood supply from the hepatic
artery. Hepatic tumors are divided into hypovascular
or hypervascular, relative to the vascularity of the
liver. Most benign and malignant hepatic masses are
relatively hypovascular, and are best detected as low

signal intensity lesions against a background of a
maximally enhanced liver achieved during the portal
venous dominant phase of hepatic enhancement.
Tumors that are considered hypervascular receive a
rich hepatic arterial flow and enhance to a similar
degree as liver parenchyma, and may not be detected
on the portal venous dominant phase of enhancement.
They are best detected on the arterial dominant phase
of enhancement.
Extracellular gadolinium chelates are considered
essential for standard MR imaging examination of
the liver, especially in imaging of hepatocellular carcinoma (HCC) or hypervascular malignant tumors
[24]. Gadolinium is administered as a rapid intravenous (IV) bolus, and imaging is performed with a
T1-weighted breath-hold GRE acquisition that is repeated dynamically [8,24]. After IV administration,
gadolinium is distributed rapidly from the vascular
compartment into the extracellular space prior to
renal excretion [25], similar to iodinated contrast
agents used for CT. Gadolinium has paramagnetic
complexes that reduce T1 relaxivity to a much greater
extent than T2 relaxivity, resulting in tissue signal
intensity increases on T1-weighted images [25]. The
time distribution of gadolinium chelates is similar to
iodine agents. In general, this means that maximum
contrast effect generally occurs at 20 and 60 seconds
(arterial and portal venous phase imaging) following
IV contrast administration.
Conventional T1 spin echo sequences acquired
after contrast administration have not been found to
increase the rate of lesion detection. This is because
peak enhancement of the liver occurs long before the
T1 images are acquired. However, rapid breath-hold
techniques are quite suitable for contrast-enhanced
imaging. In particular, General Electrics (GEs) fast
multiplanar spoiled gradient recalled (FMPSPGR)
sequence and Siemens 2D fast low-angle shot
(FLASH) sequence have been demonstrated to have
excellent utility when used with IV contrast [4,14]. For
both sequences, the TR is proportional to the number
of slices acquired; therefore, the TR and number of
slices are manipulated to acquire as many images as
possible during the estimated duration of the patients
ability to hold his or her breath. These sequences are
T1 weighted and may be used with fat suppression. In
general, ideal imaging parameters include flip angle of
70 to 80, TR of 110 to 140, TE minimum, and
receiver bandwidth of F 16 kHz. More recently 3D
GRE sequences such as volume-interpolated breathhold examination [14] or FMPSPGR have been successfully utilized. These provide thin sections, fat
suppression, and high signal-to-noise ratio. They have
53
been valuable in generating high-quality MR angiograms, allowing for accurate depiction of the vascular
anatomy of the liver.
RES contrast agents
Ion oxide particulate agentsformerly termed
superparamagnetic iron oxide (SPIO)are currently
identified as ferumoxides. Ferumoxides (marketed in
the United States under the tradename Feridex, Berlex
Laboratories Inc., Wayne, NJ) are taken up by cells of
the reticuloendothelial system in the liver, spleen, and
bone marrow. Ferumoxide is administered IV in
100 mL D5W over approximately 30 minutes. The
agent is distributed relatively rapidly to the liver,
spleen, and bone marrow. The agent is contraindicated
in patients with hemochromatosis or hypersensitivity
to iron. Other than idiosyncratic reactions, the other
possible side effect is back pain, although this is
infrequent (4% of patients) [26]. This side effect
begins 5 7 minutes after contrast agent administration, is probably related to the particulate agents, and
is not specific to ferumoxides. If the patient complains
of back pain, the infusion should be stopped immediately. The pain will gradually resolve over 10 to
15 minutes. In our experience, we have been able to
restart and finish administration of the ferumoxide,
and acquire the necessary sequences.
Ferumoxides are T2* contrast agents, so the largest signal change is on T2-weighted and T2*weighted images. The liver shows decreased signal
intensity, as does the spleen and marrow. The normal
liver takes up the agent, resulting in increased contrast-to-noise ratio between tumor and normal liver
[6,7,26]. On T1 images, there is typically some
circulating contrast agent and blood vessels show
increased signal intensity. Our current protocol
includes T1-weighted breath-hold GRE images of
the liver, and FSE T2-weighted images, which require
about 15 minutes. The patient is then removed from
the scanner, and the contrast agent is administered.
After contrast administration, the same pulse sequences are repeated again. Hepatocellular lesions, such as
adenoma or focal nodular hyperplasia (FNH), contain
reticuloendothelial cells, so they will behave similar
to the liver, with decreased signal on T2-weighted
images [27]. Focal lesions that do not contain reticuloendothelial cells maintain nearly the same signal
intensity before and after the administration of contrast material, but are rendered relatively higher in
signal due to a loss of signal in background liver.
Moderate or poorly differentiated HCC do not change
signal after ferumoxide; thus, they are easily detected
(Fig. 1). Well-differentiated HCC, however, may
contain reticuloendothelial cells and may show signal
54
Fig. 1. Patient with cirrhosis and poorly differentiated HCC. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds)
shows slightly hyperintense tumor in the right lobe (arrow). (B) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds)
30 minutes after ferumoxide injection shows drop in signal intensity of the liver and spleen, but no enhancement of the tumor.
This results in increased tumor conspicuity.
loss after ferumoxide, and are difficult to detect.

Similarly, FNH frequently shows significant signal
loss after ferumoxide, similar to normal liver (Fig. 2).
Ferumoxides may be most useful clinically in
evaluating liver metastases [28,29] and in the
evaluation of intrahepatic cholangiocarcinoma [30].
Studies have shown that ferumoxide-enhanced
T2-weighted images perform comparable to CT
arterial portography in the demonstration of liver
metastases [6]. In contrast to nonspecific extracellular gadolinium chelates [5], there is no requirement that the patient breath hold or that the MR
imaging machine be able to perform breath-hold
sequences; therefore, image quality in general is
more reproducible [26].
Smaller particulate agents are known as ultrasmall particles iron oxide (USPIO) [26], and these
have a more prolonged intravascular half-life than
do the larger particle agents. In a dilute intravascular
phase, USPIO possess T1 effects that are similar to
the vascular phase effects of T1 agents. Therefore,
they can provide additional characterization information. Additionally, USPIO can be administered as a
rapid IV bolus in a very small volume, making them
more convenient to use and minimizing potential
back pain.
Hepatocyte-selective contrast agents
An agent commonly utilized is marketed currently
under the name Teslascan (Nycomed Inc., Princeton,
Fig. 2. Patient with FNH. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) showing subtle lesion (arrow) that
is minimally hyperintense to liver. (B) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) 30 minutes after
ferumoxide injection showing homogenous tumor enhancement. FNH frequently shows significant signal intensity loss after
ferumoxide, usually more than do other primary liver tumors. This is due to the presence of reticuloendothelial cells within the
tumor, resulting in increased contrast uptake.
NJ). The contrast agent contains manganese, which is

paramagnetic (ie, imaging properties that are similar
to gadolinium-based agents). Mangafodipir is a hepatocellular contrast agentit is taken up directly by
the hepatocytes. It is administered intravenously over
approximately 1 to 3 minutes (slow push). Maximum
contrast effect has been reported at approximately
15 minutes. Because it is a T1 contrast agent, the
most effect pulse sequence appears to be T1-weighted
breath-hold GRE.
The utility of this agent is still being explored.
The agent should result in increased T1 contrast
between normal liver (increased signal intensity)
and metastatic disease (decreased signal intensity)
on T1 images (Fig. 3) [31]. The degree of enhancement of HCC correlates with histological differentiation, potentially providing a noninvasive index of
tumor biology (Fig. 4). Teslascan may potentially
have utility in identifying hepatocellular lesions
55
such as FNH, adenoma (Fig. 5), and well-differentiated HCC.

Because hepatocyte-selective contrast agents are
eliminated in part through the biliary system, they
provide an evaluation of hepatocyte function and the
biliary tree [32]. Unlike nonspecific extracellular
gadolinium chelates, imaging with Teslascan does
not require breath-hold sequences. This agent appears
to be safe and well tolerated.
Combined perfusional and hepatocyte-selective
contrast agents
Contrast agents with combined perfusional and
hepatocyte-selective functions possess both early
perfusional information via renal elimination and
later hepatocyte-selective information via hepatic
excretion. These include gadolinium benzyloxypropionictetraacetate and gadolinium ethoxybenzyl
diethylenetriminepentaacetic acid [33]. These agents
Fig. 3. Patient with colon cancer and liver metastasis. (A) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds,
70 flip angle) shows a subtle slightly hypointense lesion in segment 4b (arrow). (B) Axial, FSE, T2-weighted image (TR/TE
3500/100 milliseconds) shows the lesion as hyperintense related to liver parenchyma. (C) Axial, T1-weighted, FMPSPGR image
(TR/TE 110/4.4 milliseconds, 70 flip angle) 3 minutes after Teslasan injection showing increased lesion conspicuity. Metastases
do not contain hepatocytes, and therefore do not enhance following Teslascan administration.
56
Fig. 4. Patient with exophytic fibrolamellar HCC. (A) Axial, T1-weighted FMPSPGR image (TR/TE 110/4.4 milliseconds, 70
flip angle) shows large exophytic mass connected to segment 2 of the left lobe (arrow). (B) Axial, T1-weighted, FMPSPGR
image (TR/TE 110/4.4 milliseconds, 70 flip angle) 3 minutes after Teslasan injection shows mass enhancing similar to the same
degree as the liver. This confirms that the mass is hepatic in origin, and not a gastrointestinal stromal tumor.
combine the lesion detection and characterization

information provided by nonspecific extracellular
gadolinium chelates with the hepatocyte information
provided by hepatocyte-selective contrast agents.
In the detection of metastasis, arterial and portal
venous phase images are similar to gadolinium or
iodine; thus, most metastases remain hypointense on
T1 weighted images. Hypervascular metastases
enhance in the arterial phase. On delayed (10
20 minutes) images, metastases are hypointense to
liver on T1 weighted images (Fig. 6). HCC are usually
hyperenhancing on arterial phase and hypointense on
portal venous phase, similar to gadolinium or iodine.
Well-differentiated HCC that may contain functioning
hepatocytes can enhance on delayed images. Similarly,

FNH may demonstrate delayed enhancement due to
the presence of functioning hepatocytes (Fig. 7).
MR imaging features of liver tumors

Benign liver lesions
Incidental benign liver lesions are very common
[34], and confident diagnosis is possible in the
majority of cases using MR imaging, without the
need for surgery, biopsy, or even follow-up imaging.
These lesions are discussed below.
Fig. 5. Patient with two large adenomas. (A) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip angle)
shows a large hemorrhagic adenoma (arrow). A subtle nonhemorrhagic lesion is identified in the right lobe (curved arrow). (B)
Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip angle) 3 minutes after Teslasan injection shows
homogenous enhancement of the nonhemorrhagic mass, likely due to the presence of hepatocytes.
57
Fig. 6. Patient with metastatic colorectal cancer. (A) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70
flip angle) shows an ill-defined hypointense lesion in segment 4b (arrow). (B) Axial, T1-weighted, FMPSPGR image (TR/TE
110/4.4 milliseconds, 70 flip angle) 60 seconds after Eovist injection (Diagnostic Imaging Berlex Laboratories Inc., Wayne, NJ)
shows minimal heterogeneous enhancement of the same lesion. (C) Axial, T1-weighted, FMPSPGR image (TR/TE 110/
4.4 milliseconds, 70 flip angle) 20 minutes after Eovist injection shows better definition of tumor contour and increased lesion
conspicuity. Metastases do not enhance on delayed images, and therefore can be differentiated from other hepatic tumors.
Hepatic cysts
Simple cysts are derived from biliary endothelium. They contain thin serous fluid, not bile, and
are lined by a single layer of epithelium. They occur
in 5% to 14% of the population and are more
prevalent in women [34]. Simple cysts are usually
solitary to few in number, but innumerable cysts
may occur in autosomal-dominant or recessive polycystic disease.
Like other cysts, the criteria for diagnosis of
hepatic cyst include sharp margins with no definable
walls. On MR, cysts show internal signals on T1 and
T2 sequences that follow cerebrospinal fluid (CSF).
After contrast administration, cysts do not show
enhancement. For small lesions that are less than or

equal to 15 mm, a small increase in signal on
contrast-enhanced scans is probably due to partial
volume averaging with normal enhancing liver.
The differential diagnosis of a hepatic cyst includes cystic metastases such as leiomyosarcoma
and hemangioma. Note that so-called cystic metastases are usually pseudocystic and tend to be internally inhomogeneous.
Cystlike hepatic lesions may result from trauma,
infection, or neoplasm. In most of these instances,
clinical evaluation is straightforward and imaging
will reveal some complexity of the cystic mass, such
as thick septa, mural nodularity, or fluid-debris levels.
58
Fig. 7. Patient with FNH. (A) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip angle) 25 seconds
after Eovist injection shows two hyperenhancing liver lesions (arrows). (B) Axial, T1-weighted, FMPSPGR image (TR/TE 110/
4.4 milliseconds, 70 flip angle) 20 minutes after Eovist injection shows some enhancement of both lesions, probably related to
impaired excretion of the contrast agent.
Hemangioma
Hemangioma is the most common benign hepatic
neoplasm, occurring in 5% to 20% of the population
[34]. Women are affected more than are men and the
lesions are commonly found in a subcapsular location
in the right lobe of the liver. Most lesions grow
slowly and are uncommonly diagnosed in children.
Even large and exophytic hemangiomas are usually
asymptomatic, and spontaneous hemorrhage or other
complications are extremely rare. Hemangiomas consist of large endothelial-lined vascular spaces separated by fibrous septa. They can range in size from a
few millimeters to more than 20 cm. Although
usually solitary, multiple hemangiomas are commonly encountered.
On MR, hemangiomas have a characteristic
appearance of a 1-cm to 3-cm lesion that is extremely bright (light bulb) on T2 images (Fig. 8)
[12,35,36]. Lesions larger than about 3 cm tend to
be less homogeneously bright, due to internal fibrous
septa or scarring. The tumor margin is well defined,
and may be somewhat lobulated. No peritumoral
edema is present. Long TE images (TE 120
180 milliseconds) will demonstrate that hemangiomas have a signal similar to CSF, whereas most
malignant tumors will not be as bright as CSF. There
are two differential diagnostic considerations for
hemangioma: a cyst and a hypervascular hepatic
metastasis. Differentiation of hemangioma from a
cyst is unimportant because both are benign. The
signal characteristics of some malignant tumors,
however, will overlap with that of hemangioma [37]
because of similar T2 values (most hemangiomas
have calculated T2 values greater than 88 milliseconds at 1.5 T). For example, metastatic leiomyo-
sarcoma in the liver may frequently (75%) have an

appearance that is similar to hemangioma.
Because of this overlap, characteristics at contrastenhanced MR have been used to further differentiate
hemangiomas from metastases. Suspected hemangiomas should be evaluated using a dynamic breath-held
acquisition over 5 to 15 minutes, in a manner similar
to that performed for dynamic CT evaluation. Following rapid contrast injection, a GRE T1-weighted
sequence is acquired in 20 to 30 seconds (FMPSPGR,
2D FLASH). This is repeated once per minute, until
the lesion has filled in completely or nearly completely, or until a diagnosis can be rendered. Using
this protocol, hemangiomas demonstrate hyperintense
peripheral nodular enhancement, with either complete
or partial filling in 5 to 15 minutes [36]. The specificity of these criteria approaches 100%.
FNH
FNH may occur in 2% to 5% of the population
[34]. It is rarely symptomatic, although lesions may
grow to more than 10 cm in diameter. Lesions are
solitary in about 70% of patients. Multiple lesions of
FNH have been described in patients with certain
brain neoplasms and vascular malformations of various organs [38]. FNH is a nonencapsulated firm
nodule of normal hepatocytes with a distinct central
scar and thin radiating fibrous septa containing
Kupffer cells and primitive bile ductules. Intratumoral
calcification, fat, hemorrhage, and necrosis are extremely rare.
FNH should be differentiated from adenoma,
because the two lesions occur in a similar patient
population. An additional important differential consideration is fibrolamellar HCC. FNH has no malig-
59
Fig. 8. Patient with large hepatic hemagioma. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) shows a large
hyperintense mass in the left lobe (arrow). (B) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip
angle) 60 seconds after gadolinium injection shows peripheral nodular enhancement, that continues to fill in on delayed
(3 minutes) images (C). These features are characteristic of hemangioma.
nant potential and is unlikely to hemorrhage; adenomas are often resected because of the potential for
hemorrhage and malignant degeneration.
On MR images, FNH is typically isointense to liver
on T1 images, and slightly hyperintense on T2 images.
On T1 images, an internal low signal area that corresponds to a central scar is present in more than 50% of
cases [39,40]. The scar is usually hyperintense to liver
on T2 images, but may be hypointense. A central scar,
best seen on T1 images, is highly suggestive of the
diagnosis. Other than the central scar, the lesions are
homogeneous on both T1 and T2 images.
FNH lesions are highly vascular. Unlike the
central scars in hemangioma, the central scar in
FNH is vascular and is surrounded by collagenous
fibers. Following contrast administration, because of
their prominent vascularity, FNH lesions tend to
enhance early, and may be isointense to liver as early
as 1 to 2 minutes after contrast injection, except for
the central scar (Figs. 2, 7, 9). After administration of
gadopentetate dimeglumine, central scars are frequently seen in FNH lesions.

Hepatocellular adenoma
Adenoma is a rare hepatic neoplasm that occurs
almost exclusively in the liver that has abnormal
metabolism due to exogenous steroids or, less commonly, congenital abnormalities of carbohydrate
metabolism. Over 90% of cases occur in women of
reproductive age, and lesions may regress with withdrawal of oral contraceptives or other steroids. Adenoma is of clinical interest because of its tendency
toward spontaneous rupture and hemorrhage. Rare
cases of malignant degeneration are also recorded,
particularly in patients with multiple or large adenomas. Adenomas are composed of cords of hepatocytes that occasionally form bile. Adenomas lack bile
ductules, portal tracts, and hepatic veins, which result
in necrosis and hemorrhage. Excessive glycogen and
lipids are characteristic histologic findings.
60
Fig. 9. Patient with FNH. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) shows a slightly hyperintense
lesion in the right lobe (arrow). (B) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip angle)
20 seconds after gadolinium injection shows homogenous hyperenhancement of the lesion. (C) Axial, T1-weighted, FMPSPGR
image (TR/TE 110/4.4 milliseconds, 70 flip angle) 60 seconds after gadolinium injection shows minimal residual tumor
enhancement. Imaging in this phase only may miss hypervascular liver lesions.
Adenomas are sharply defined masses with smooth

borders. Unlike FNH, adenomas are usually heterogeneous due to the presence of hemorrhage, necrosis,
or fat. Patients may present with hepatomegaly and
acute onset of abdominal pain secondary to spontaneous hemorrhage. Hepatic adenomas may infrequently
undergo malignant transformation [41]. In addition, as
they enlarge, they may be the source of spontaneous
intra-abdominal hemorrhage. For these reasons, surgeons will typically resect solitary liver adenomas.
The MR appearance of adenomas is somewhat
variable, but the typical tumor demonstrates slight
increased signal on T1 images (probably due to fat
content) and moderate increased signal on T2 images.
As opposed to FNH, they are more frequently heterogeneous in their internal signal. This is, in part,
due to the tendency of this tumor to undergo spontaneous internal hemorrhage with areas of calcification and fibrosis (see Fig. 5). The appearance of
adenoma tends to overlap with that of HCC.
Biliary cystadenoma
This tumor presents typically as a symptomatic
large hepatic mass characterized by large cystic spaces
surrounded and separated by thickened, nodular,
enhancing walls. Although imaging features might
overlap with a pyogenic or parasitic abscess, this is
usually not a clinical dilemma. MR imaging studies
clearly demonstrate cystadenoma as an aggressive
and complex neoplasm (Fig. 10) [42]. These are
known to undergo malignant transformation and there
is no imaging technique capable of distinguishing
between benign and malignant cystadenomas.
Malignant liver lesions
Hepatic metastasis
Metatstatic disease to the liver is approximately
20 times more common than is primary hepatic neoplasms. The presence of metastatic disease to the liver
is a prime determinant of survival. Prognosis is inver-
61
Fig. 10. Patient with biliary cystadenoma. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) shows a large
hyperintense cyst (arrow), with associated ductal dilatation in both lobes, particularly the left lobe (curved arrow). (B) Axial,
T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip angle) 60 seconds after gadolinium injection shows cystic
mass with enhancement of a mural nodule (arrow).
sely proportional not only to the presence of metastases, but also to the number and volume of metastases. The most common primary tumors to
metastasize to the liver are colon, breast, lung, pancreas, melanoma and sarcoma. Metastases in the liver
usually appear as focal, discrete lesions, but diffuse
infiltrative involvement of the liver may be seen in
breast cancer or lymphoma, and rarely with colon or
lung primaries.
For small lesions that are less than or equal to
15 mm in size, benign tumors (hemangioma, cyst)
must be considered in the differential diagnosis of
metastatic disease, even if multiple lesions are found.
In one series of 209 patients with a known primary
malignancy, 51% of patients with lesions that were

less than 15 mm in size were diagnosed as having
benign lesions [43].
The most common appearance of metastatic disease in the liver by MR is rounded or focal lesions
with decreased signal on T1 images and moderately
increased signal on T2 images (see Fig. 3) [44].
Malignant tumors tend to have margins that are not
sharply defined, and the signal intensity on T2 images
is usually not as bright as CSF. Tumors may be single
or multiple.
Hypervascular metastases are those with an
abundant blood supply, typically greater than that of
normal liver (Fig. 11). These tumors include chorio-
Fig. 11. Patient with metastatic carcinoid. (A) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip
angle) 20 seconds after gadolinium injection shows innumerable rim-enhancing lesions. (B) Axial, T1-weighted, FMPSPGR
image (TR/TE 110/4.4 milliseconds, 70 flip angle) 60 seconds after gadolinium injection. The lesions are less conspicuous on
this portal venous phase of enhancement.
62
carcinoma, renal cell carcinoma, thyroid carcinoma,

breast carcinoma, melanoma, carcinoid tumor, and
islet cell tumor. In general, these tumors may have
hyperintense signal on T2 images, and so potentially
could be mistaken for hemangioma or cyst [45]. In
addition, most of these tumors have an increased
tendency to have intratumoral hemorrhage. This
results in increased signal on T1 images as well as
T2 images. Melanoma, due to the T1 shortening
effect of melanin, shows increased signal on T1
image even if hemorrhage is not present.
HCC
The incidence of HCC is rising in the United States
and has almost doubled over the past 20 years
[46 49]. This rise is caused in part by the epidemic
of hepatitis C virus, which can lead to both cirrhosis
and HCC. Cirrhosis due to hepatitis C causes 70%
of the HCC cases in Japan and 30% to 50% of the
HCC cases in the United States, although 40% of the
HCC seen in North America occurs in noncirrhotic
livers [49].
The primary goal of imaging is lesion detection
and characterization. Because of the propensity of
HCC to invade vessels, vascular involvement should
be carefully evaluated. Patients with chronic liver
diseaseparticularly cirrhosis, hepatics B or C, and
hemochromatosisare at risk for HCC.
Presentation in the United States is often with
pain, weight loss, and elevated alpha-fetoprotein.
There are three patterns of HCC in the liver: a solitary
mass, a dominant mass with smaller satellite lesions
(ie, multifocal HCC, 20%), and diffuse involvement.
Diffuse involvement is more common in patients with
cirrhosis. HCCs are highly vascular lesions and may
infrequently present with spontaneous hemorrhage.

Most commonly, lesions are well circumscribed with
the appearance of a well-defined capsule.
On MR, the T1 appearance of HCC ranges from
hypointense to slightly hyperintense, depending on
fat content, copper deposition within the tumor, and
the degree of differentiation. On T2 images, most
HCC demonstrate increased signal compared to the
surrounding liver, although the tumors tend to be
inhomogeneous (Figs. 1, 12). Contrast-enhanced MR
imaging may play a role in characterizing and
detecting small HCC in patients with underlying
cirrhosis. In these cases, the T1 and T2 of the tumors
may not be substantially different from that of the
surrounding liver, or the underlying liver heterogeneity may make the tumor difficult to detect.
Additionally, in patients with cirrhosis and ascites,
significant motion artifacts obscure the underlying
liver parenchyma. In these cases, imaging following
rapid IV injection of gadolinium may increase lesion conspicuity. Typically, a fast breath-held GRE
acquisition is used to acquire several sets of images
immediately following the end of the contrast injection. The hepatic artery supplies HCC, whereas
the normal liver receives its dominant (80%) blood
supply from the portal vein. Early imaging at 30 to
40 seconds after contrast injection results primarily
in a hepatic arterial phase image of the liver, and
some tumorsparticularly small, well-differentiated
tumorsare better visualized during this phase of
hepatic enhancement.
Fibrolamellar HCC is a distinctive type of HCC
that occurs in younger patients (mean age of
20 years). Although it is a malignant lesion, the prognosis is better than that for typical HCC, with 25% of
Fig. 12. Patient with HCC. (A) Axial, FSE, T2-weighted image (TR/TE 3500/100 milliseconds) shows a slightly hyperintense
heterogeneous mass in segment 4 (arrow). (B) Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip
angle) 60 seconds after gadolinium injection shows hypoenhancing mass abutting, but not invading, left portal vein (arrow).
63
difficult to distinguish tumor from bile duct and

vessels. For this reason, contrast-enhanced MR is
recommended using a T1-weighted GRE sequence.
On T1 images, the bile ducts are dark (fluid). The
tumor demonstrates late enhancement following
gadolinium [50], so that differentiation between
tumor and bile ducts is readily recognized (Fig. 13).
Ferumoxides have also been proven to be valuable in
imaging patients with intrahepatic cholangiocarcinoma [30]. Vascular invasion and portal nodes should
be carefully searched for because these findings
preclude resection of the tumor. Solitary peripheral
masses appear similar to HCC or hepatic metastases.
Fig. 13. Patient with cholangiocarcinoma. Axial, T1-weighted, FMPSPGR image (TR/TE 110/4.4 milliseconds, 70 flip
angle) 3 minutes after gadolinium injection shows delayed
enhancement of the central mass (arrow), probably due to
increased fibrous tissue. An internal biliary stent is also
identified (curved arrow).
patients having resectable lesions. Alpha-fetoprotein

levels are usually not elevated. Fibrolamellar HCC is
typically a well-circumscribed lesion that is hypointense on T1 images and hyperintense on T2 images
(see Fig. 4). A central scar may be present. Central
calcifications are present in one third of lesions. The
differential diagnosis includes adenoma or FNH.
Cholangiocarcinoma
Cholangiocarcinomas are adenocarcinomas of the
biliary tree. These lesions are hypovascular with
abundant stroma. Approximately 10% to 25% are
intrahepatic. Predisposing factors include primary
sclerosing cholangitis, infestation with Clonorchis
sinensis, and Carolis disease.
Klatskin tumor is a small stricturing cholangiocarcinoma arising at the junction of the left and right
hepatic ducts. These lesions produce bilobar biliary
duct obstruction and are nearly always unresectable.
The differential diagnosis includes central metastasis
or possibly lymphoma. Peripheral cholangiocarcinoma may occasionally be resectable when it does
not involve the inferior vena cava or caudate lobe.
Ten percent of patients with sclerosing cholangitis
may have associated cholangiocarcinoma.
The most prominent feature with central cholangiocarcinoma on MR is usually intrahepatic biliary
duct dilatation. Morphologic changes may occur late
in the disease process, with atrophy of the left lobe of
the liver compared with the right lobe. The left-sided
hepatic ducts may be more dilated than are those in
the right lobe. Central tumors tend to show high
signal intensity on T2 images near the porta hepatis,
but because of associated bile duct dilatation, it is
Summary
The article reviews the current MR imaging
techniques commonly utilized for imaging liver
tumors. Breath-hold T1-weighted GRE sequences,
FSE T2-weighted sequences, and properly timed
contrast-enhanced 3D SGE are important for lesion
characterization. New liver-specific contrast agents
improve lesion detection and are useful in lesion characterization.
References
[1] Rofsky NM, Weinreb JC, Bosniak MA, et al. Renal
lesion characterization with gadolinium-enhanced MR
imaging: efficacy and safety in patients with renal insufficiency. Radiology 1991;180(1):85 9.
[2] Bluemke DA, Paulson EK, Choti MA, et al. Detection
of hepatic lesions in candidates for surgery: comparison
of ferumoxides-enhanced MR imaging and dual phase
helical CT. AJR Am J Roentgenol 2000;175(6):
1653 8.
[3] de Lang EE, Mugler 3rd JP, Gay SB, et al. Focal liver
disease: comparison of breath-hold T1 weighted MPGRE MR imaging and contrast-enhanced CTlesion
detection, localization, and characterization. Radiology
1996;200:465 73.
[4] Semelka RC, Willms AB, Brown MA, et al. Comparison of breath-hold T1-weighted MR sequences for
imaging of the liver. J Magn Reson Imaging 1994;
4(6):759 65.
[5] Semelka RC, Worawattanakul S, Kelekis NL, et al.
Liver lesion detection, characterization, and effect on
patient management: comparison of single-phase spiral
CT and current MR techniques. J Magn Reson Imaging
1997;7(6):1040 7.
[6] Seneterre E, Taourel P, Bouvier Y, et al. Detection of
hepatic metastases: ferumoxides-enhanced MR imaging versus unenhanced MR imaging and CT during
arterial portography. Radiology 1996;200(3):785 92.
64
[7] Ward J, Naik KS, Guthrie JA, et al. Hepatic lesion

detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dualphase CT by using alternative-free response receiver
operating characteristic analysis. Radiology 1999;
210(2):459 66.
[8] Yamashita Y, Mitsuzaki K, Yi T, et al. Small hepatocellular carcinoma in patients with chronic liver damage: prospective comparison of detection with dynamic
MR imaging and helical CT of the whole liver. Radiology 1996;200(1):79 84.
[9] Bartolozzi C, Lencioni R, Donati F, et al. Abdominal
MR: liver and pancreas. Eur Radiol 1999;9:1496 512.
[10] Siegelman ES, Outwater EK. MR imaging techniques
of the liver. Radiol Clin North Am 1998;36:263 86.
[11] Semelka RC, Shoenut JP, Kroeker MA, et al. Focal liver
disease: comparison of dynamic contrast-enhanced CT
and T2-weighted fat-suppressed, FLASH, and dynamic
gadolinium-enhanced MR imaging at 1.5 T. Radiology
1992;184:687 94.
[12] Hamm B, Thoeni RF, Gould RG, et al. Focal liver
lesions: characterization with nonenhanced and dynamic contrast material-enhanced MR imaging. Radiology 1994;190(2):417 23.
[13] Semelka RC, Simm FC, Recht M, et al. T1-weighted
sequences for MR imaging of the liver: comparison of
three techniques for single-breath, whole-volume
acquisition at 1.0 and 1.5 T. Radiology 1991;180:
629 35.
[14] Rofsky NM, Lee VS, Laub G, et al. Abdominal MR
imaging with a volumetric interpolated breath-hold examination. Radiology 1999;212(3):876 84.
[15] Lee VS, Lavelle MT, Rofsky NM, et al. Hepatic MR
imaging with a dynamic contrast-enhanced isotropic
volumetric interpolated breath-hold examination: feasibility, reproducibility, and technical quality. Radiology 2000;215(2):365 72.
[16] Rydberg JN, Lomas DJ, Coakley KJ, et al. Comparison
of breath-hold fast spin-echo and conventional spinecho pulse sequences for T2-weighted MR imaging
of liver lesions. Radiology 1995;194(2):431 7.
[17] Kanematsu M, Hoshi H, Itoh K, et al. Focal hepatic
lesion detection: comparison of four fat-suppressed
T2-weighted MR imaging pulse sequences. Radiology
1999;211:363 71.
[18] Keogan MT, Spritzer CE, Paulson EK, et al. Liver MR
imaging: comparison of respiratory triggered fast spin
echo with T2-weighted spin-echo and inversion recovery. Abdom Imaging 1996;21:433 9.
[19] Lu DS, Saini S, Hahn PF, et al. T2-weighted MR imaging of the upper part of the abdomen: should fat suppression be used routinely? AJR Am J Roentgenol
1994;162(5):1095 100.
[20] Schwartz LH, Seltzer SE, Tempany CM, et al. Prospective comparison of T2-weighted fast spin-echo, with
and without fat suppression, and conventional spinecho pulse sequences in the upper abdomen. Radiology
1993;189(2):411 6.
[21] Kiefer B, Grassner J, Hausmann R. Image acquisition
in a second with Half-Fourier single shot turbo spin

echo. J Magn Reson Imaging 1994;4P:486 9.
[22] Morrin MM, Rofsky NM. Techniques for liver MR
imaging. Magn Reson Imaging Clin N Am 2001;
9(4):675 96.
[23] Tang Y, Yamashita Y, Takahashi M. Ultrafast
T2-weighted imaging of the abdomen and pelvis: use
of single shot fast spin-echo imaging. J Magn Reson
Imaging 1998;8:384 90.
[24] Low RN, Francis IR, Sigeti JS, et al. Abdominal MR
imaging: comparison of T2-weighted fast and conventional spin-echo, and contrast-enhanced fast multiplanar spoiled gradient-recalled imaging. Radiology
1993;186(3):803 11.
[25] Peterson MS, Baron RL, Murakami T. Hepatic malignancies: usefulness of acquisition of multiple arterial
and portal venous phase images at dynamic gadolinium-enhanced MR imaging. Radiology 1996;201(2):
337 45.
[26] Ros PR, Freeny PC, Harms SE, et al. Hepatic MR
imaging with ferumoxides: a multicenter clinical trial
of the safety and efficacy in the detection of focal
hepatic lesions. Radiology 1995;196(2):481 8.
[27] Vogl TJ, Hammerstingl R, Schwarz W, et al. Magnetic
resonance imaging of focal liver lesions. Comparison
of the superparamagnetic iron oxide resovist versus
gadolinium-DTPA in the same patient. Invest Radiol
1996;31(11):696 708.
[28] Bellin MF, Zaim S, Auberton E, et al. Liver metastases:
safety and efficiency of detection with superparamagnetic iron oxide in MR imaging. Radiology 1994;193:
657 63.
[29] Shamsi K, Balzer T, Saini S, et al. Superparamagnetic
iron oxide particles (SH U 555 A): evaluation of efficacy in three doses for hepatic MR imaging. Radiology
1998;206(2):365 71.
[30] Braga HJ, Imam K, Bluemke DA. MR imaging of
intrahepatic cholangiocarcinoma: use of ferumoxides
for lesion localization and extension. AJR Am J Roentgenol 2001;177(1):11 4.
[31] Wang C, Ahlstrom H, Ekholm S, et al. Diagnostic efficacy of MnDPDP in MR imaging of the liver. A phase
III multicentre study. Acta Radiol 1997;38(4 Pt 2):
643 9.
[32] Mitchell DG, Alam F. Mangafodipir trisodium: effects
on T2- and T1-weighted MR cholangiography. J Magn
[33] Caudana R, Morana G, Pirovano GP, et al. Focal malignant hepatic lesions: MR imaging enhanced with
gadolinium benzyloxypropionictetra-acetate (BOPTA)preliminary results of phase II clinical application. Radiology 1996;199(2):513 20.
[34] Karhunen PJ. Benign hepatic tumours and tumour like
conditions in men. J Clin Pathol 1986;39(2):183 8.
[35] Semelka RC, Helmberger TK. Contrast agents for MR
imaging of the liver. Radiology 2001;218(1):27 38.
[36] Whitney WS, Herfkens RJ, Jeffrey RB, et al. Dynamic
breath-hold multiplanar spoiled gradient-recalled MR
imaging with gadolinium enhancement for differentiat-
[37]
[38]
[39]
[40]
[41]
[42]
[43]
ing hepatic hemangiomas from malignancies at 1.5 T.

Radiology 1993;189(3):863 70.
Lombardo DM, Baker ME, Spritzer CE, et al. Hepatic
hemangiomas vs metastases: MR differentiation at
1.5 T. AJR Am J Roentgenol 1990;155(1):55 9.
Wanless IR, Albrecht S, Bilbao J, et al. Multiple focal
nodular hyperplasia of the liver associated with vascular malformations of various organs and neoplasia of
the brain: a new syndrome. Mod Pathol 1989;2(5):
456 62.
Mathieu D, Rahmouni A, Anglade MC, et al. Focal
nodular hyperplasia of the liver: assessment with contrast-enhanced TurboFLASH MR imaging. Radiology
1991;180(1):25 30.
Vilgrain V, Flejou JF, Arrive L, et al. Focal nodular
hyperplasia of the liver: MR imaging and pathologic
correlation in 37 patients. Radiology 1992;184(3):
699 703.
Foster JH, Berman MM. The malignant transformation of liver cell adenomas. Arch Surg 1994;129(7):
712 7.
Gabata T, Kadoya M, Matsui O, et al. Biliary cystadenoma with mesenchymal stroma of the liver:
correlation between unusual MR appearance and
pathologic findings. J Magn Reson Imaging 1998;8:
503 4.
Jones EC, Chezmar JL, Nelson RC, et al. The fre-
[44]
[45]
[46]
[47]
[48]
[49]
[50]
65
quency and significance of small (less than or equal

to 15 mm) hepatic lesions detected by CT. AJR Am J
Roentgenol 1992;158(3):535 9.
Imam K, Bluemke DA. MR imaging in the evaluation
of hepatic metastases. Magn Reson Imaging Clin N
Am 2000;8(4):741 56.
Larson RE, Semelka RC, Bagley AS, et al. Hypervascular malignant liver lesions: comparison of various
MR imaging pulse sequences and dynamic CT. Radiology 1994;192(2):393 9.
Brown JJ, Naylor MJ, Yagan N. Imaging of hepatic
cirrhosis. Radiology 1997;202(1):1 16.
El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med
1999;340(10):745 50.
Ince N, Wands JR. The increasing incidence of hepatocellular carcinoma. N Engl J Med 1999;340(10):
798 9.
Nzeako UC, Goodman ZD, Ishak KG. Hepatocellular
carcinoma in cirrhotic and noncirrhotic livers. A clinico-histopathologic study of 804 North American patients. Am J Clin Pathol 1996;105(1):65 75.
Low RN, Semelka RC, Worawattanakul S, et al. Extrahepatic abdominal imaging in patients with malignancy: comparison of MR imaging and helical CT in
164 patients. J Magn Reson Imaging 2000;12(2):
269 77.
MR imaging of diffuse liver disease

Ioana-Maria Danet, MD, Richard C. Semelka, MD*, Larissa Braga, MD
Department of Radiology, University of North Carolina at Chapel Hill, CB# 7510, Chapel Hill, NC 27599 7510, USA
The recent development of rapid-acquisition techniques with excellent image quality and tissue-specific contrast agents has rendered MR imaging the
most accurate imaging modality for the evaluation of
liver disease. In many instances, with the appropriate
combination of sequences, MR imaging can diagnose
and characterize diffuse liver disease accurately so
that the need for invasive procedures is obviated. MR
imaging plays an important role in the evaluation of
complications and follow-up of diffuse hepatic disease and may have prognostic and therapeutic implications. Diffuse liver disease by definition refers to
any destructive or infiltrative process that involves
the liver parenchyma, the biliary ducts, or the vascular structures leading to liver dysfunction. Diffuse
liver disease includes a variety of conditions, such as
infectious and inflammatory diseases, storage and
metabolic disorders, neoplastic diseases, and vascular
diseases. Cirrhosis, a complex and slowly destructive
process of various etiologies, is considered as a
distinct entity. This article discusses abdominal MR
imaging techniques and findings on the most common causes of diffuse liver disease with an emphasis
on MR imaging patterns and pathologic correlation.
MR imaging technique
The current standard MR imaging examination
of the liver includes a combination of T1- and
Ioana-Maria Danet, MD, is funded by a grant from the

Royal College of Physicians and Surgeons of Canada.
Larissa Braga, MD, is supported by CAPES, Brazil.
E-mail address: richsem@med.unc.edu (R.C. Semelka).
T2-weighted sequences followed by the acquisition

of dynamic contrast-enhanced T1-weighted images
after injection of a rapid bolus of gadolinium chelate.
By using the appropriate combination of sequences, a
comprehensive evaluation of the liver can be
achieved with approximately 15 minutes of table time
[1 3]. The most commonly used T1-weighted
sequence in high field systems ( > 1 T) is breathhold
spoiled gradient echo (SGE) sequence. This sequence
provides excellent T1-weighted contrast, and permits
entire image acquisition of the entire liver in one
breathhold. Moreover, breathing artifacts and magnetic susceptibility artifacts are minimized [4].
Dynamic contrast-enhanced breathhold acquisitions
have become the cornerstone of MR imaging of the
liver and may be the single most important acquisition for lesion detection and characterization. Because the central phase encoding steps of the
multisection acquisition are acquired over 6 seconds
for each individual section and the entire data set,
precise timing of data acquisition is achieved during
the arterial-dominant and venous phases of enhancement. This almost simultaneous acquisition of the
bulk signal has allowed a temporal resolution superior to that currently reached with contrast-enhanced
CT. A newly developed three-dimensional gradient
echo sequence, also referred to as volume interpolated breathhold examination, is currently under
investigation as for its application as a T1-weighted
sequence [5]. This sequence provides high spatial
resolution, multiplanar display with high definition of
hepatic blood vessels, and acquisition of thinner
sections that may allow detection of smaller lesions.
For the nonbreathhold T1-weighted sequence, a
breathing-independent magnetization-prepared gradient echo sequence can acquire a single slice in less
than 2 seconds [4]. This sequence is relatively
PII: S 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 0 - X
68
I.-M. Danet et al / Radiol Clin N Am 41 (2003) 6787
insensitive to motion artifacts. On T1-weighted images, high fluid content lesions, such as cysts or
hemangiomas, hypovascular lesions, or lesions with
a high fibrotic content are moderately low in signal
intensity [6]. Hemorrhagic lesions, lesions with high
protein content, or lesions containing melanin are of
high signal intensity.
The most commonly performed T2-weighted
sequences are breathing-averaged conventional spin
echo or faster sequences, such as echo-train spin echo
sequences or a modified turbo spin echo sequence
termed half Fourier single-shot turbo spin echo
(HASTE). The combination of T2-weighted images
with fat suppression improves the detection of focal
liver lesions within a fatty infiltrated liver and the
detection of lymphadenopathy in the porta hepatis.
Moreover, fat-suppressed T2-weighted images diminish phase artifacts from respiration. On T2-weighted
images, lesions with increased fluid content are of
high signal intensity, whereas the presence of fibrotic
tissue or increased iron content is low in signal
intensity [6]. T2-weighted echo train spin echo
sequences, such as the HASTE sequence, provide an
excellent anatomic display of the common bile duct on
coronal plane images.
The development of phased-array or multiarray
torso coils has improved the signal-to-noise ratio
permitting faster acquisition of thinner sections with
improved spatial resolution.
The current standard protocol as used in the
authors center includes fat-suppressed T2-weighted
HASTE, turbo short tau inversion recovery, precontrast T1-weighted SGE, opposed-phase SGE, and
postgadolinium dynamic imaging in the arterial-dominant phase, the portal phase, and the interstitial phase.
Cirrhosis
Cirrhosis is a slowly progressive condition in
which irreversible damage to liver parenchyma
caused by inflammation and necrosis is followed by
fibrous scarring and development of regenerative
nodules. Although a variety of hepatic disorders
may lead to cirrhosis, in North America the most
common causes include alcoholism and viral hepatitis
[7]. MR imaging can evaluate the morphologic
changes of cirrhotic livers, characterize focal hepatic
masses, and provide a noninvasive evaluation of
intrahepatic and extrahepatic vessels. In the authors
center, MR imaging is used routinely for the followup of patients with cirrhosis and for the evaluation of
liver transplant candidates.
MR imaging may become a valuable imaging

modality from the therapeutic point of view, because
it can diagnose disease at an early stage when there
is more to offer clinically. Early in the evolution
of cirrhosis, MR imaging can depict subtle morphologic changes, such as fine strands of fibrosis and
enlargement of the hilar periportal space. Heavily
T1-weighted images, such as out-of-phase SGE
images, are very sensitive to subtle fibrotic changes
because fibrous tissue is more conspicuous on short
TE sequences. Early fibrotic changes can be appreciated as a lace-like hypointense fine network on short
TE SGE images. Linear enhancement in the interstitial
phase of enhancement reflects the distribution of
gadolinium in the large extracellular space of the
fibrotic septal tissue. Enlargement of the hilar periportal space, defined as the space anterior to the right
portal vein, is considered to be an early manifestation
of cirrhosis. The enlargement of the hilar periportal
space is caused by atrophy of segment four and was
visible in 98% of patients with early cirrhosis [8]. Early
detection of cirrhosis is of clinical importance because
these patients may benefit from new treatments, such
as interferon for hepatitis C [9].
As morphologic changes in cirrhotic livers progress, there is marked atrophy of the medial segment
of the left lobe and the right lobe with sparing or
hypertrophy of the central segments of the left lobe
and the caudate lobe. Expansion of the interlobar
fissure has been defined as the expanded gallbladder
fossa sign. This sign is frequently seen and has been
shown to have a high specificity and positive predictive value (98% for each) for diagnosing cirrhosis
[10]. In the advanced stage of cirrhosis, atrophy of
the lateral segments of the left lobe and the caudate
lobe may occur. Fibrotic septa of various thickness
and confluent fibrosis become conspicuous as a low
signal intensity network on T1-weighted images. On
T2-weighted images, newly formed inflammatory
fibrous tissue is of high signal intensity. Fibrotic
tissue shows negligible enhancement immediately
following gadolinium administration and progression
in the late interstitial phase (Fig. 1) [11].
The combination of atrophic and hypertrophic
changes combined with the formation of regenerative
nodules and fibrosis may lead to a complex nodular
parenchymal distortion. The International Working
Party nomenclature [12] of nodular lesions in cirrhotic
livers includes regenerative nodules and low- and
high-grade dysplastic nodules. Most small hepatocellular carcinoma (HCC), defined as lesions with a
diameter less than 2 cm, are thought to develop in a
stepwise pathway from regenerative nodules and
subsequently from dysplastic nodules. Careful evalua-
69
Fig. 1. Cirrhosis. T2-weighted single shot-echotrain spin echo (A), spoiled gradient echo (SGE) (B), out-of-phase (C), immediate
postgadolinium SGE (D), 45-second postgadolinium SGE (E), and 90-second postgadolinium fat-suppressed SGE (F) images.
The liver is diminutive in size and demonstrates nodular and irregular contour with distorted anatomy. The hepatic parenchyma is
heterogeneous in appearance with extensive linear fibrosis. The fibrous stroma is best shown on out-of-phase image (C) as lowsignal reticular strands and on the late postgadolinium fat-suppressed SGE (F) as enhancing tissue.
tion and understanding of the MR imaging patterns of

different nodular lesions in cirrhotic livers is important for the early detection of small HCC. The diagnosis of small HCC is of clinical importance because
these patients may benefit from minimal invasive
treatments, such as thermal ablation, cryotherapy,
laser-guided therapy, ethanol injection, or arterial
chemoembolization. A combination of T1-weighted,

T2-weighted, and serial gadolinium-enhanced SGE
images can accurately define and characterize most
lesions in cirrhotic liver. Regenerative nodules are
clusters of newly formed hepatocytes surrounded by a
network of fibrotic septa [12]. Although micronodular
and macronodular regenerative changes have been
70
described on the basis of pathologic gross inspection

and related to specific etiologies, such as viral or
alcohol-induced cirrhosis, a more dynamic process
with conversion from micronodular to macronodular
form is thought to take place. Most cirrhotic livers
demonstrate a mixed composition [13] and a conver-
sion ratio from the micronodular form to the macronodular form has been found in 90% of cases in
10 years [14].
MR imaging can demonstrate regenerative nodules
with greater sensitivity than any other imaging modality. Regenerative nodules are hypointense to hyper-
Fig. 2. Cirrhosis with regenerative nodules. Spoiled gradient echo (SGE) (A), out-of-phase SGE (B), immediate postgadolinium
SGE (C), 45-second postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E) images. The liver is
diminutive in size and shows irregular nodular contours consistent with cirrhosis. Multiple-sized siderotic nodules are
appreciated throughout the liver and exhibit low signal on T1-weighted (A) images and negligible enhancement after
administration of contrast (C E) compatible with regenerative nodules.
intense on T1-weighted images, and isointense to

hypointense on T2-weighted images relative to the
surrounding inflammatory fibrous septa [15]. On postgadolinium SGE images they are of low signal intensity because hepatic parenchyma enhances greater
than the regenerative nodules (Fig. 2). Iron accumulation can be found in up to 25% of regenerative nodules,
71
facilitating their identification as low signal intensity

on T2-weighted images and gradient echo images,
which are particularly sensitive to magnetic field
inhomogeneities (see Fig. 2) [16,17].
Dysplastic nodules represent a premalignant intermediary stage in the pathway of hepatocarcinogenesis
and may represent up to 25% of cirrhotic nodules [13].
Fig. 3. Cirrhosis with high-grade dysplastic nodule and early hepatocellular carcinoma. Echo-train short tau inversion recovery (A),
spoiled gradient echo (SGE) (B), immediate postgadolinium SGE (C), 45-second postgadolinium SGE (D), and 90-second
postgadolinium fat-suppressed SGE (E). There is 1-cm lesion in the anterior part of the right lobe that is not evident on T2-weighted
(A) image, demonstrates a slightly high signal on T1-weighted (B) image, and displays an intense enhancement on the immediate
postcontrast image (C) that persists on late image (E), consistent with severe dysplastic nodule.
72
Dysplastic nodules are classified by the International

Working Party as low or high grade with regard to the
presence of dysplasia and atypia. On MR imaging, the
combination of hyperintensity on T1-weighted images
and hypointensity on T2-weighted images has been
described as a characteristic appearance for dysplastic
nodules. Because there is much overlap in the T1- and
T2-weighted appearance between regenerative, dysplastic, and early HCC, however, their differentiation
on the basis of these sequences is limited (Fig. 3). As
the degree of malignancy increases in the pathway of
hepatocarcinogenesis, portal blood supply is expected
to decrease, whereas the arterial blood supply increases [18].
Fig. 4. Hepatocellular carcinoma. Echo-train short tau inversion recovery (A), spoiled gradient echo (SGE) (B), immediate
postgadolinium SGE (C), 45-second postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E) images.
There is a lesion in the left hepatic lobe that demonstrates slightly heterogeneous high signal on T2-weighted images (A), slightly
low signal on T1-weighted images (B), intense uniform enhancement on the immediate postgadolinium image (C), and fades to
hypointensity by 45 (D) and 90 (E) seconds. Note the enhancement of a pseudocapsule, more pronounced on late image (E). This
lesion is compatible with hepatocellular carcinoma.
Although enhancement in the early arterial-dominant phase may be present in dysplastic nodules,
intense early enhancement is a distinctive feature of
HCC. The presence of early arterial enhancement
with rapid washout during the portal phase should
be regarded as highly suspicious for the presence of
a small HCC (Fig. 4). As HCC develops in the
dysplastic nodule, hyperintensity on T2-weighted
images becomes apparent. The nodule within a nodule sign has been described for a hypointense
dysplastic nodule harboring a small hyperintense
HCC on T2-weighted images [19]. Imaging features
that suggest the presence of HCC are (1) lesion size
greater than 3 cm, (2) hyperintensity on T2-weighted
images, (3) intense enhancement in the arterial-dominant phase, (4) late tumor washout, (5) the presence
of a capsule, and (6) a rapid rate of growth on followup studies. The presence of a capsule is relatively
common in HCC, whereas it is seldom found in
other malignant liver lesions. MR imaging is the
73
most accurate imaging modality in demonstrating

tumor capsule. The capsule appears as low signal
intensity on T1-weighted images and demonstrates
enhancement on the interstitial phase of gadoliniumenhanced images.
Cirrhotic livers may demonstrate heterogeneous
enhancement during the hepatic arterial phase. This
can be explained by repetitive episodes of acute
inflammation and hepatocellular damage or by focal
reduction of portal blood flow with compensatory
increase in the arterial flow [9,20,21]. On MR
imaging, acute on chronic liver inflammation appears
as low signal intensity on T1-weighted images and
high signal intensity on T2-weighted images with
transient increased enhancement on immediate postgadolinium images (Fig. 5) [9]. Distinction from
tumor is usually not problematic because these
regions are small and have ill-defined irregular margins. When patchy areas of enhancement are large,
however, distinction from diffuse HCC may be prob-
Fig. 5. Acute or chronic hepatitis. Echo-train short tau inversion recovery (A), spoiled gradient echo (SGE) (B), immediate
postgadolinium SGE (C), and 90-second postgadolinium fat-suppressed SGE (D) images in a patient with a chronic hepatitis.
The left lobe and part of the right lobe demonstrate a well-delimitated region with slightly high signal on T2- (A) and T1(B) weighted images, increased enhancement on immediate postcontrast (D) images, and persistence of the signal on late image
consistent with acute or chronic hepatitis.
74
Fig. 6. Cirrhosis and peritoneal enhancement. Transverse 90-second postgadolinium fat-suppressed spoiled gradient echo (A,B)
images in a patient with history of cirrhosis. A mild linear peritoneal enhancement (arrow) consistent with microvarices in the
peritoneal lining caused by portal hypertension is demonstrated.
lematic. In this particular setting, ancillary findings,

such as tumor thrombus and high levels of alpha
fetoprotein, are highly suggestive of the presence of
diffuse HCC.
Postsinusoidal portal hypertension is a complication of advanced cirrhosis and includes splenomegaly,
development of collateral circulation, varices, and
ascites (Fig. 6). With the development of substantial
portosystemic shunting, the volume of blood flow to
the liver is reduced and so is the size of the portal vein.
If thrombosis of the portal vein occurs, a centripetal
hepatic flow can be maintained if collaterals within the
porta hepatis develop, also referred to as cavernous
transformation. MR imaging demonstrates particularly well the peritoneal varices on the fat-suppressed
late gadolinium-enhanced SGE images as small
enhancing serpiginous foci along the peritoneal surfaces (Fig. 7). MR imaging is more sensitive than
angiography, endoscopy, or contrast-enhanced CT
imaging for detecting varices. Gadolinium-enhanced
water excitation SGE images can further accentuate
the high signal in vessels produced by gadolinium
enhancement, because water excitation possesses
some time-of-flight properties resulting in high signal
in flowing vessels. Because MR imaging is particularly sensitive to soft tissue contrast, early mesenteric
and omental edema are particularly well demonstrated
as a mild infiltrative haze on T2-weighted images.
Cirrhosis induced by primary sclerosing cholangitis shows a distinct morphologic pattern. In primary
sclerosing cholangitis there is progressive fibrosis
and segmental narrowing of the intrahepatic or extrahepatic bile ducts with final progression to biliary
cirrhosis. MR imaging is a useful noninvasive
method for the evaluation of the biliary tree and for
the demonstration of the hepatic morphologic

changes induced by primary sclerosing cholangitis.
MR cholangiography has been shown to be accurate
in the detection and localization of primary sclerosing
cholangitis [22] with a specificity and sensitivity of
85% to 88% and 92% to 97%, respectively. In the
authors experience, an MR imaging pattern relatively common and distinctive of primary sclerosing
cholangitis induced cirrhosis is characterized by
central macroregenerative nodules and peripheral
liver atrophy. The central regenerative nodules may
reach impressive dimensions and may contribute to
the periportal compression and obstruction of the bile
ducts (Fig. 8). MR imaging features of primary
sclerosing cholangitis include (1) thickening (50%)
and enhancement (67%) of the bile ducts wall [23];
(2) wedge-shaped areas of increased signal intensity
on T2-weighted images (72%) [24]; (3) high signal
intensity along the porta hepatis secondary to periportal edema (40%); and (4) segmental or wedgeshaped areas of high signal intensity on T1-weighted
images not related to focal fatty infiltration (21%).
Patchy peripheral or segmental areas of increased
parenchymal enhancement on arterial-dominant
phase images were identified in 56% of patients
and suggest an increased arterial flow with decreased
portal flow [23].
Infectious and inflammatory disorders

Hepatitis
Viral hepatitis is the most common cause of acute
inflammatory liver disease. The most common caus-
75
Fig. 7. Cirrhosis and omental hypertrophy. Echo-train short tau inversion recovery (A), spoiled gradient echo (SGE) (B),
immediate postgadolinium SGE (C), and 90-second postgadolinium fat-suppressed SGE (D) images. The liver is small, nodular
in contour, and demonstrates a reticular heterogeneous enhancement pattern consistent with cirrhosis. Note also the omental
hypertrophy (arrow), recanalization of umbilical vein, and ascites.
ative agents are hepatitis B, C, or D viruses. Other

causes of acute hepatitis include toxic hepatitis, such
as alcohol abuse and radiation-induced hepatitis. In
acute viral hepatitis, the degree of parenchymal
damage is variable and the major pathologic findings
are inflammatory infiltrates and hepatocyte necrosis.
Hepatocyte regeneration is encountered in the healing regenerative phase. Acute hepatitis is diagnosed
by clinical and serologic studies. Imaging studies
may be performed if the clinical picture is worrisome. On MR imaging, acute hepatitis may show
nonspecific findings, such as heterogeneous signal
intensity most apparent on T2-weighted images and a
heterogeneous pattern of enhancement on the arterial-dominant phase SGE images (Fig. 9). High signal
intensity periportal edema may be identified on T2weighted images. In both acute and chronic hepatitis
adenopathy in the periportal space may be the only
abnormality identified.
Chronic hepatitis is characterized by persistent

inflammatory infiltrates and necrosis associated with
abnormal liver enzymes values. Severe chronic hepatitis may progress to cirrhosis and there is extensive
overlapping between the MR imaging features of
chronic hepatitis and those of early cirrhosis. The
MR imaging appearance of the liver may provide
complementary information on the degree of histologic activity of the disease, and may help to monitor
the response to treatment. On T2-weighted images, a
homogeneous or heterogeneous increase in signal
intensity has been described and reflects the presence
of inflammation or necrosis of liver parenchyma
[21,25,26]. In a recent study, Semelka et al [11]
described the enhancement patterns in chronic hepatitis with histopathologic correlations. Early patchy
enhancement in patients with chronic hepatitis was
shown to be associated with the presence of significant parenchymal inflammatory reaction, indicating
76
Fig. 8. Primary sclerosing cholangitis. T2-weighted SS-ETSE fat-suppressed (A), spoiled gradient echo (SGE) (B), immediate
postgadolinium SGE (C), 45-second postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E) images.
The liver shows distorted anatomy with massive enlargement of caudate lobe and atrophy of the peripheral liver. The signal
intensity is heterogeneous on T2- and T1-weighted images with multiple macronodules and fibrotic bands. Note also, that there is
ductal dilatation in the peripheral liver because of the central hypertrophy that contributes to periportal compression or
obstruction of bile ducts. These findings are consistent with cirrhosis caused by primary sclerosing cholangitis.
77
Fig. 9. Acute hepatitis postchemotherapy. Spoiled gradient echo (SGE) (A), immediate postgadolinium SGE (B), 45-second
postgadolinium SGE (C), and 90-second postgadolinium fat-suppressed SGE (D) images in a patient in chemotherapy treatment.
The liver is enlarged and demonstrates a patchy heterogeneous transient enhancement after administration of contrast consistent
with acute hepatitis.
either concurrent or recent hepatocellular damage

(see Fig. 5) [11]. Absence of early patchy enhancement correlated with low inflammatory reaction or
hepatocyte necrosis. Regional modifications of liver
blood supply secondary to the underlying chronic
liver disease may also contribute to create a wedgeshaped patchy enhancement. Late linear enhancement reflects the presence of fibrosis.
Storage and metabolic diseases

Fatty liver
Fatty liver or steatosis is defined as the accumulation of fat, mainly triglycerides, within the hepatocytes. Diffuse steatosis is a common condition and
autopsy series found steatotic changes in up to 7% of
nonobese patients and in 29% of obese patients
[27,28]. A variety of clinical disorders are associated
with steatosis including obesity, malnutrition, diabetes
mellitus, steroid use, alcoholic liver disease, and
hepatitis. Diffuse fatty infiltration can be uniform or
patchy. Although steatosis can be demonstrated on CT

or sonography, MR imaging is considered to be the
most accurate imaging modality in the evaluation of
fatty infiltration of the liver. A combination of in-phase
and out-of-phase T1-weighted images accurately characterizes fatty liver infiltration and differentiates focal
fatty infiltration from potential hepatic lesions. Moreover, out-of-phase techniques render hepatic lesions
more conspicuous because the contrast between the
liver and hepatic lesions is maximized.
Breathhold SGE images are particularly suitable
for the implementation of in-phase and out-of-phase
images using identical parameters with the exception
of TE. With complementary in-phase and out-of-phase
SGE images, separation of signals from fat and water is
made possible on the basis of the difference in the
resonant frequencies between fat and water [29]. The
difference in contrast between in-phase and out-ofphase images is greater at short TE. At 1.5 T the lowest
in-phase echo time is 4.2 milliseconds and the lowest
out-of-phase echo time is 2.1 milliseconds. By setting
the TE, in-phase or out-of-phase images can be
obtained and compared. In fatty livers, opposing
78
signals from fat and water cancel each other and there
is loss of signal intensity on out-of-phase images by
comparison with in-phase images [30]. The maximal
signal loss is observed when fat content approaches
50% of the voxel in liver. The spleen can be used as the
organ of reference for the signal loss, and when the fat
and water are present in equal proportion in the voxel
the liver appears darker relative to the spleen. For

lesser amounts of fatty infiltration (15%) the signal
intensity of liver appears nearly equal to that of the
spleen on out-of-phase images and is slightly more
intense on in-phase images.
Focal steatosis and focal sparing may represent a
diagnostic challenge particularly in the clinical set-
Fig. 10. Fatty infiltration with focal sparing. Echo-train short tau inversion recovery (A), spoiled gradient echo (SGE) (B), out-ofphase SGE (C), immediate postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E) images.
Homogeneous signal of the liver is present on the T1-weighted image (B). On the out-of-phase image (C) the liver drops in signal
but a persistent focus of higher signal is still appreciated in segment 8/5. Note that in 90-second image (E) the same region is
identified as higher signal. This reflects a fat-suppression effect of the remainder of the liver rather than a gadolinium-enhanced
effect of the regions of normal liver.
ting of malignancy. In most cases both focal sparing

and focal steatosis have characteristic locations, such
as the gallbladder fossa, the medial segment near the
falciform ligament, the subcapsular region, and the
porta hepatis. These foci represent foci of liver
tissue that may experience differing blood supply
and differing fat deposition than the remainder of
the liver. Focal fatty infiltration can be demonstrated
as a region of signal loss in out-of-phase images.
Of all hepatic lesions that contain fat, the most
susceptible to resemble a focal fatty infiltration
is adenoma, because it may have a uniform fat
content. The differentiation of a distinct lesion from
focal fatty infiltration relies on lesion morphology,
enhancement pattern, and signal characteristics on
T1- and T2-weighted images. Entirely fatty lesions,
such as lipoma or angiomyolipoma, do not show a
drop in signal but may show a phase-cancellation
artifact [31]. Focal fatty sparing presents as a
hypointense region on in-phase images and can
mimic metastasis or other hepatic lesions in the
setting of fatty liver. On out-of-phase images focal
sparing shows an increase in signal compared with
the low signal liver parenchyma (Fig. 10). Imaging
features of focal fatty liver or focal sparing include
location, wedge-shape configuration with angulated
margins, lack of mass effect, coursing of normal
79
vascular structures through the area of concern,

and enhancement indistinguishable from that of
normal liver. In contrast focal lesions almost invariably show a different enhancement pattern than
background liver (eg, adenomas show intense homogeneous enhancement immediately following gadolinium administration).
Iron deposition
When excess iron is present either because of
increased intestinal absorption or administration by
intravenous blood transfusion, there is abnormal iron
deposition in various tissues, also referred to as
hemochromatosis. MR imaging is the most sensitive
and specific imaging modality for the demonstration
of iron overload and also for the follow-up of patients
under treatment. MR imaging quantitative analysis of
iron stores is feasible, although it is not generally
performed. The authors in their clinical practice
categorize iron deposition as mild, moderate, or
severe. MR imaging can be used for the evaluation
of the distribution of iron deposition or as a primary
diagnostic method in the investigation of a symptomatic patient. When placed within a magnetic field,
iron acquires strong magnetization and induces field
inhomogeneity for the neighboring water protons. As
Fig. 11. Primary hemochromatosis. Echo-train short tau inversion recovery (A), in-phase spoiled gradient echo (SGE) (B), and
out-of-phase SGE (C) images. The liver and pancreas are low signal intensity on T2- (A) and T1-weighted (B) images consistent
with iron deposition. The spleen is relatively normal in signal intensity.
80
a result, these protons lose phase coherence and

cause a reduction in signal intensity. The paramagnetic effect is particularly conspicuous on T2- and
T2*-weighted images. T2*-weighted gradient
recalled echo sequences lack the 180-refocusing
pulse and are more sensitive to magnetic field inhomogeneities, and have a greater sensitivity for mild
degrees of iron infiltration than spin echo T2weighted image sequences. Qualitative evaluation
of the iron deposition within liver parenchyma can
be done by comparing the signal intensity of liver
with that of adjacent paraspinal muscle that is unaffected by iron overload. If the signal intensity of
liver is less than that of paraspinal muscles, iron
accumulation should be considered.
Primary hemochromatosis
Primary hemochromatosis is a common genetic
disorder among the white population in the United
States [32]. Excessive gastrointestinal absorption of
dietary iron results in progressive deposition in the
parenchymal cells of the liver, heart, and pancreas,
with subsequent organ dysfunction. Early in the
course of the disease, the liver is the first organ that
shows evidence of increased iron content [33]. At
this stage, treatment by phlebotomy usually results in
normal life expectancy. Long-standing complications
are cirrhosis and HCC, which are found in up to 35%
of patients. Other manifestations of the disease
include diabetes mellitus, hyperpigmentation, and
cardiomyopathy with arrhythmias and congestive
heart failure. In primary hemochromatosis there is a
lack of selective iron accumulation in the reticuloendothelial system of the spleen and the spleen
maintains high signal intensity, whereas the liver
and pancreas show a decrease in signal intensity
(Fig. 11). The presence of iron accumulation in the
pancreas is an indicator of advanced disease and
correlates with irreversible changes of cirrhosis in

the liver (see Fig. 11).
Erythrogenic hemochromatosis is characterized by
iron overload in the parenchymal cells of the liver and
the pancreas secondary to ineffective erythropoiesis,
such as thalassemia major. The MR imaging appearance is similar to that of primary hemochromatosis
because the absorbed iron is preferentially deposited
in the parenchymal cells. Although quantitative MR
imaging techniques are not widely available, a qualitative MR imaging evaluation of iron overload is
recommended for the monitoring and follow-up of
patients under therapy.
Secondary hemochromatosis
Transfusional overload is the most common cause
of secondary hemochromatosis. Iron accumulates in
the reticuloendothelial cells of the liver, spleen, and
bone marrow and does not tend to produce clinically
significant organ dysfunction. The pancreas is usually
spared in transfusional overload, whereas the spleen
shows accumulation of iron.
A mildly increased hepatocyte iron overload can
be found in patients with cirrhosis, the cause of
which is not well understood. It may be related to
anemia, pancreatic insufficiency, or decreased transferrin synthesis.
Mucopolysaccharidoses
The mucopolysaccharidoses are a group of inherited disorders characterized by incomplete degradation and storage of acid mucopolysaccharides.
Mucopolysaccharides accumulate in many organ systems resulting in hepatosplenomegaly, skeletal
deformities, valvular and vascular deposits, and central nervous system abnormalities. The diagnosis is
made by means of clinical and radiologic manifes-
Fig. 12. Storage disease. T2-weighted SS-ETSE (A) and spoiled gradient echo (B) images in a patient with a history of
mucopolysacharidosis demonstrate an enlargement of the liver. No other abnormalities are observed.
tations and by finding increased urinary excretion of

mucopolysaccharides and deficiency of specific
enzymes [34]. On MR imaging hepatosplenomegaly
is commonly observed (Fig. 12), but more specific
features are yet to be elucidated.
Diffuse neoplastic diseases

Diffuse HCC
The diffuse form of HCC is rare and may represent a diagnostic challenge within a cirrhotic liver.
Acute on chronic hepatitis and confluent early hepatic
fibrosis may mimic diffuse HCC. The most common
appearance of diffuse HCC is that of diffuse mottled
punctuated hyperintensities on T2-weighted images
and mottled intense enhancement in the arterialdominant phase of enhancement. Unlike HCC, fibrosis shows marked enhancement in the interstitial
phase of enhancement. Fibrotic tissue may enhance
early but to a lesser degree than the surrounding
hepatic parenchyma, and on delayed images fibrosis
appears hyperintense relative to liver parenchyma.
Areas of acute inflammation superimposed on cirrhotic liver may demonstrate signal and enhancement
characteristics similar to those of diffuse HCC. Findings in favor of HCC are a nodular appearance with
mass effect on liver contour and vessels. The presence of thrombosis in the portal vein is highly
associated with diffuse HCC. Moreover, high levels
of serum a-fetoprotein are invariably associated with
diffuse HCC [6].
Metastatic disease
Diffuse metastatic involvement of the liver has
been described with breast carcinoma, melanoma,
small cell carcinoma, and pancreatic neoplasms. Confluent segmental or lobar involvement with metastasis
often appears moderately hypointense on T1-weighted
images, mildly hyperintense on T2-weighted images,
and usually shows increased enhancement on postgadolinium images and fades over time (Fig. 13). On MR
imaging, diffuse liver involvement by metastatic
implants can mimic the appearance of a grossly
cirrhotic liver. Diffuse metastatic infiltration of the
liver with desmoplastic reaction and cirrhotic-like
morphologic changes has been described in patients
with breast carcinoma who had undergone chemotherapy [35]. The pathogenesis of the cirrhotic-like appearance is still unclear, but likely represents extensive
fibrosis of innumerable small metastases secondary to
chemotherapy [35]. Diffuse metastatic involvement of
81
the liver should be considered in the proper clinical

setting and in the presence of MR imaging suggestive
of cirrhosis.
Lymphoma
Secondary liver involvement by either Hodgkins
or non-Hodgkins lymphoma is common in stage IV
of the disease. Diffuse liver involvement is rare and is
characterized by infiltration of neoplastic cells without significant morphologic changes. On MR
imaging the liver parenchyma may show slightly
increased signal intensity [36]. Tumoral infiltration
may occur along the portal tracts. Periportal tracking
may be visualized as moderately high signal intensity
on T2-weighted fat-suppressed images and on
venous-phase gadolinium-enhanced images.
Hepatic vascular disorders

Heterogeneous patterns of enhancement may be
encountered in various conditions, such as inflammation of the liver parenchyma or of the biliary tree,
cirrhosis, tumoral infiltration, vascular obstruction, or
vascular imbalance without any definable cause. In
this latter entity, a heterogeneous patchy enhancement
may reflect the imbalance between hepatic arterial
and portal venous blood with increased arterial blood
supply to areas that show increased early enhancement. Moreover, vascular obstructions of the portal
vein, the hepatic veins, or the hepatic arteries may
cause a heterogeneous aspect of the liver parenchyma
on T1- and T2-weighted images and a heterogeneous
pattern of enhancement. MR imaging has the advantage to offer noninvasive evaluation of the hepatic
vasculature, and assessment of the vascular patency
must be included as part of a comprehensive MR
imaging evaluation of the liver.
Portal vein thrombosis

Thrombosis of the portal vein can occur at the
extrahepatic or intrahepatic level and may be partial
or complete. Common causes include pancreatitis,
ascending cholangitis, peritoneal sepsis, lymphadenopathy, neoplasm, and stasis associated with cirrhosis. MR imaging has been reported to be more
sensitive than CT in detecting portal vein thrombosis
and collateral vessels [37]. The thrombus can be
bland or tumoral. Bland thrombi have low signal
intensity on T2-weighted images and on SGE images
and do not enhance with gadolinium. On the other
82
Fig. 13. Liver metastases. Coronal T2-weighted SS-ETSE (A), spoiled gradient echo (SGE) (B), out-of-phase SGE (C),
immediate postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E) images. There are multiple small
lesions scattered throughout hepatic parenchyma. These lesions are slightly high signal on T2-weighted (A) image, low signal on
T1-weighed (B) image, and demonstrate a uniform ring enhancement on immediate postgadolinium image (D) that fades over
time (E) with negligible central enhancement consistent with metastases.
83
Fig. 14. Subacute Budd-Chiari syndrome. T2-weighted SS-ETSE fat-suppressed (A), spoiled gradient echo (SGE) (B),
immediate postgadolinium SGE (C), 45-second postgadolinium SGE (D), and 90-second postgadolinium fat-suppressed SGE (E)
images. The liver is enlarged, irregular in contour, and with hypertrophy of caudate lobe. The signal of the peripheral liver is
slightly high on T2-weighted (A) images and slightly low on T1-weighted (B) images. Postadministration of contrast there is an
increased heterogeneous enhancement in the peripheral liver in comparison with central liver. Over time the central area becomes
more homogeneous with the remainder of the liver.
84
hand, tumor thrombi are of slightly higher signal

intensity on T2-weighted images, of soft tissue
intensity on SGE images, and show enhancement
with gadolinium administration. Tumor thrombi usually occur in the presence of HCC but may occasionally be seen with other neoplastic or metastatic
processes. Extrinsic compression of the portal vein
can be seen at the intrahepatic or extrahepatic level
and is usually related to the presence of lymphadenopathy, tumor, or nodular parenchymal hypertrophy
as seen in cirrhosis secondary to primary sclerosing
cholangitis. Cholangiocarcinoma has been described
to have a propensity to cause portal vein compression, causing a wedge-shaped area of high signal
intensity on T2-weighted images and increased early
arterial enhancement [38]. When the main portal vein
is thrombosed a centripetal blood flow can be supplied to the liver by a network of portal collateral
vessels, a process referred to as cavernous transformation [39]. On MR imaging, thrombosis of the
intrahepatic portion of the portal vein may produce
segmental or lobar wedge-shaped areas of increased
enhancement in the arterial-dominant phase. Areas of

early transient increased enhancement correspond to
areas with decreased portal flow and resulting
increased compensatory hepatic arterial supply secondary to autoregulatory mechanisms. Often simple
portal vein thrombosis does not result in signal
change on T2-weighted images. On later images a
homogeneous liver enhancement is observed as the
concentration of gadolinium in the hepatic arteries
and portal vein equilibrates. Other findings associated
with long-term thrombosis of the portal vein are
atrophy of the subtended liver segments with compensatory hypertrophy of the other segments.
Budd-Chiari syndrome
Budd-Chiari syndrome (BCS) is an uncommon
cause of obstruction of the hepatic venous outflow
leading to progressive hepatic failure. The venous
obstruction can be located at the intrahepatic level or
at the level of the inferior vena cava (IVC). Most
cases of BCS are idiopathic. Other known causes of
Fig. 15. Congestive liver. Spoiled gradient echo (SGE) (A), immediate postgadolinium SGE (B), 45-second postgadolinium SGE
(C), and 90-second postgadolinium SGE (C) images. The liver is enlarged with a dilated inferior vena cava. A mosaic pattern of
enhancement is demonstrated in 45-second images (C) and became less evident in 90-second images (D).
BCS are hematologic disorders, pregnancy or the use

of oral contraceptives, intravascular webs or membranes, or tumor thrombi extending from the IVC
[40]. Imaging findings in BCS range from direct
visualization of the intraluminal thrombus to late
morphologic changes of the liver parenchyma. In
some cases, the obstruction of the hepatic veins
may be segmental or subsegmental and the demonstration of patent main hepatic veins does not exclude
BCS. A combination of black blood and bright blood
technique was shown to be effective in demonstrating
the presence or absence of flow in the hepatic veins
and IVC [41].
Common morphologic changes in long-standing
BCS are atrophy of the peripheral liver parenchyma,
which experiences severe venous obstruction, and
hypertrophy of the caudate lobe and of the central
portions of the liver that may have a distinct venous
outflow. A recent study [41] described the MR
imaging appearance of morphologic changes and
patterns of enhancement with histologic correlations
that may be useful in differentiating the acute, subacute, and chronic BCS. MR imaging may aid in
guiding the appropriate clinical management. In the
acute syndrome, the peripheral liver demonstrates
hypointense signal intensity on T1-weighted images
and hyperintensity on T2-weighted images reflecting
the presence of edema. There is also markedly less
enhancement peripherally when compared with the
central region (Fig. 14). The increased central
enhancement is persistent on delayed images. In the
acute syndrome ascites is often present [42].
In the subacute syndrome, with the decrease in the
parenchymal pressure, the early enhancement of the
caudate lobe is less prominent than the heterogeneously increased enhancement in the peripheral
liver. Signal characteristics on T1- and T2-weighted
images are similar, however, to those observed in the
acute disease. The chronic stage is characterized by
fibrosis particularly of the peripheral liver that shows
low signal intensity on T1- and T2-weighted images.
The enhancement differences between the peripheral
and the central part of the liver are diminished and
there is massive caudate lobe hypertrophy and prominent collaterals. Regenerative and dysplastic nodules
can be observed. These nodules have been described
as having high signal intensity on T1-weighted
images and intermediate to low signal intensity on
T2-weighted images [43]. These nodules tend, however, to possess intense enhancement on the arterialdominant phase of enhancement.
In chronic BCS syndrome there is development of
portal hypertension and portosystemic collaterals.
Characteristic curvilinear intrahepatic collaterals and
85
capsular-based collaterals can be identified on gadolinium-enhanced portal phase images.

Hepatic arterial obstruction
Hepatic arterial obstruction is a rare condition
most commonly seen in the setting of liver transplantation. Embolic occlusion is another cause of
hepatic arterial occlusion. On early postgadolinium
images there is diminished enhancement of hepatic
parenchyma reflecting arterial obstruction.
Congestive heart failure
Long-term elevation of the central venous pressure, as encountered in congestive heart failure or
constrictive pericarditis, may lead to increased hepatic venous pressure and subsequently to sinusoidal
congestion and edema. A mosaic pattern of enhancement with a reticulated pattern of low signal intensity
can be seen on early dynamic contrast-enhanced MR
images (Fig. 15). Reflux of contrast from the right
atrium into the hepatic veins and suprahepatic IVC
can be identified before opacification of the portal
vein. Other ancillary findings include prominence of
the hepatic veins and IVC, ascites, cardiomegaly, and
pleural effusion.
Summary
MR imaging is able to demonstrate and distinguish the full variety of benign and malignant diffuse
liver diseases.
References
[1] Semelka RC, Balci NC, Op de Beeck B, Reinhold C.
Evaluation of a 10-minute comprehensive MR imaging
examination of the upper abdomen. Radiology 1999;
211:189 95.
[2] Semelka RC, Helmberger TK. Contrast agents for MR
imaging of the liver. Radiology 2000;218: 27 38.
[3] Semelka RC, Worawattanakul S, Kelekis NL, et al.
Liver lesion detection, characterisation and effect on
patient management: comparison of single-phase spiral
CT and current MRI techniques. J Magn Reson Imaging 1977;7:1040 7.
[4] Semelka RC, Willms AB, Brown MA, et al. Comparison of breathhold T1-weighted MR sequences for
imaging of the liver. J Magn Reson Imaging 1994;
212:876 84.
[5] Rofsky NM, Lee VS, Laub G, Pollack MA, Krinsky
86
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]

GA, Thomasson D, et al. Abdominal MR imaging
with a volumetric interpolated breathhold examination. Radiology 1999;212:876 84.
Semelka RC, Braga L, Armao D, Martin DR, et al.
Liver. In: Semelka RC, editor. Abdominal-pelvic
MRI. New York: Wiley-Liss; 2001. p. 33 317.
Gore RM. Diffuse liver disease. In: Gore RM, Levine
NS, Laufer I, editors. Textbook of gastrointestinal
radiology. Philadelphia: WB Saunders; 1994.
p. 1968 2017.
Ito K, Mitchell DG, Gabata T. Enlargement of hilar
periportal space: a sign of early cirrhosis at MR imaging. J Magn Reson Imaging 2000;11:136 40.
Bisceglie SJ, Martin P, Kasskanides C, et al. Recombinant interferon alpha therapy for chronic hepatitis C: a
randomized, double blind, placebo-controlled trial. N
Engl J Med 1989;321:1506 10.
Ito K, Mitchell DG, Gabata T, Hussain SM. Expanded
gallbladder fossa: simple MR imaging sign of cirrhosis. Radiology 1999;211:723 6.
Semelka RC, Chung JJ, Hussain SM, Marcos HB,
Woosley JT. Chronic hepatitis: correlation of early
patchy and late linear enhancement patterns on gadolinium enhanced MR images with histopathology
initial experience. J Magn Reson Imaging 2001;13:
385 91.
International Working Party. Terminology of nodular hepatocellular lesions. Hepatology 1995;22:
983 93.
Hytiroglu P, Theise NH. Differential diagnosis of hepatocellular nodular lesions. Semin Diagn Pathol 1982;
2:177 86.
Fauerholdt L, Schlichting P, Christensen E, et al. Conversion of micronodular cirrhosis into macronodular
cirrhosis. Hepatology 1983;3:928 31.
Ohtomo K, Itai Y, Ohtomo Y, Shiga J, Iio M. Regenerating nodules of the liver cirrhosis: MR imaging with
pathologic correlation. AJR Am J Roentgenol 1990;
154:505 7.
Mitchell DG, Lovett KE, Hann HW, Ehrlich S, et al.
Cirrhosis: multiobserver analysis of hepatic MR imaging findings in a heterogeneous population. J Magn
Terada T, Nakanuma Y. Survey of iron accumulative
macroregenerative nodules in cirrhotic livers. Hepatology 1989;10:851 4.
Matusi O, Kadoya M, Kameyama T, Yoshikawa J, et
al. Adenomatous hyperplastic nodules in the cirrhotic
liver: differentiation from hepatocellular carcinoma
with MR imaging. Radiology 1989;173:123 6.
Mitchel DG, Rubin R, Siegelman ES, et al. Hepatocellular carcinoma within siderotic regenerative nodules:
appearance as a nodule within a nodule on MR imaging. Radiology 1991;178:101 3.
Lehman B, Fanucci E, Gigli F, Uhnlenbrock D, Bartolozzi C. Signal suppression of normal liver tissue by
phase corrected inversion recovery: a screening technique. J Comput Assist Tomogr 1989;13:650 5.
Marti-Bonmati L, Talens A, Del Olmo J, de Val A, et
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
al. Chronic hepatitis and cirrhosis : evaluation by

means of MR imaging with histologic correlation. Radiology 1993;188:37 43.
Fulcher AS, Turner MA, Franklin KJ, Shiffman ML,
Sterling RK, Luketic VA, et al. Primary sclerosing
cholangitis: evaluation with MR cholangiographya
case-control study. Radiology 2000;215:71 80.
Ito K, Mitchell DG, Outwater EK, Blasbalg R. Primary
sclerosing cholangitis: MR imaging features. AJR Am
J Roentgenol 1999;172:1527 33.
Revelon G, Rashid A, Kawamoto S, Bluemke DA.
Primary sclerosing cholangitis: MR imaging findings
with pathologic correlation. AJR Am J Roentgenol
1999;173:1037 42.
Murakami T, Baron RL, Peterson MS. Liver necrosis
and regeneration after fulminant hepatitis: pathologic
correlation with CT and MRI findings. Radiology
1996;198:239 42.
Stark DD, Bass NM, Moss AA, et al. Nuclear magnetic
resonance imaging of experimentally induced liver disease. Radiology 1983;148:743 51.
Anderson T, Gluud C. Liver morphology in morbid
obesity: a literature study. Int J Obes Relat Metab Disord 1984;8:97 106.
Siegelman ES. MR imaging of diffuse liver disease:
hepatic fat and iron. Magn Reson Imaging Clin N
Am 1997;5:347 65.
Fishbein MH, Gardner KG, Potter CJ, et al. Introduction of fast MR imaging in the assessment of hepatic
steatosis. Magn Reson Imaging 1997;15:287 93.
Martin J, Sentis M, Puig J, et al. Comparison of inphase and opposed-phase GRE and conventional SE
MR pulse sequences in T1-weighted imaging of liver
lesions. J Comput Assist Tomogr 1996;20:890 7.
Itai Y. Focal sparing versus hepatic tumor in fatty liver.
Brandhagen DJ, Fairbanks VF, Batts KP, Thebodeau
SN. Update on hereditary hemochromatosis and the
HFE gene. Mayo Clin Proc 1999;74:917 21.
Siegelman ES, Mitchell DG, Semelka RC. Abdominal
iron deposition: metabolism. MR findings, and clinical
importance. Radiology 1996;199:13 22.
Muenzer J. Mucopolysaccharidoses. In: Behrman RE,
Kliegman RM, Jenson HB, editors. Nelson textbook
of pediatrics. 16th edition. Philadelphia: WB Saunders;
2000. p. 420 3.
Nascimento AB, Mitchell DG, Rubin R, Weaver E.
Diffuse desmoplastic breast carcinoma metastases to
the liver simulating cirrhosis at MR imaging: report
of two cases. Radiology 2001;221:117 212.
Weissleder R, Stark DD, Elizondo G, et al. MRI of
hepatic lymphoma. Magn Reson Imaging 1988;6:
675 81.
Zirinsky K, Markisz JA, Rubenstein WA, et al. MR
imaging of portal venous thrombosis: correlation with
CT and sonography. AJR Am J Roentgenol 1988;
150:283 8.
Worawattanakul S, Semelka RC, Noone TC, Calvo
BF, et al. Cholangiocarcinoma: spectrum of appearan-

ces on MR images using current techniques. Magn
Reson Imaging 1998;16:993 1003.
[39] Nakao N, Miura K, Takahashi H, Miura T, et al. Hepatic perfusion in cavernous transformation of the portal
vein: evaluation by using CT angiography. AJR Am J
Roentgenol 1989;152:985 6.
[40] Murphy FB, Steinberg HV, Shires III GT, Martin LG,
Bernardino ME. The Budd-Chiari syndrome: a review.
[41] Noone TC, Semelka RC, Siegelman ES, Balci NC, et
87
al. Budd-Chiari syndrome: spectrum of appearances of

acute, subacute and chronic disease with magnetic
resonance imaging. J Magn Reson Imaging 2000;11:
44 50.
[42] Mathieu D, Vasile N, Menu Y. Budd-Chiari syndrome:
dynamic CT. Radiology 1987;165:409 13.
[43] Soyer P, Lacheheb D, Caudron C, Levesque M. MRI of
adenomatous hyperplastic nodules of the liver in BuddChiari syndrome. J Comput Assist Tomogr 1993;17:
86 9.
MR cholangiopancreatography
Tomofumi Motohara, MDa, Richard C. Semelka, MDa,*, Till R. Bader, MDb
a
Department of Radiology, University of North Carolina, 2006 Old Clinic Building, CB #7510,
Chapel Hill, NC 27599 7510, USA
b
Department of Radiology, University of Vienna, Austria
MR cholangiopancreatography (MRCP) is still a

rapidly evolving technique, but has been already
accepted as clinically useful and is widely used to
evaluate biliary or pancreatic diseases in a noninvasive
way. This technique uses MR imaging to visualize
stationary or slow-moving fluid, such as bile, displaying them as high signal intensity. Heavily T2-weighted
sequences are generally used for MRCP with singleshot echo-train spin echo technique achieving the most
widespread use. Recent studies show that MRCP is
comparable with or more useful than other techniques,
such as ultrasound, CT, and endoscopic retrograde
cholangiopancreatography (ERCP) to study choledocholithiasis, malignant obstruction of the biliary or
pancreatic ducts, congenital anomalies, and chronic
pancreatitis [1 7]. With further improvements of
hardware and technique, MRCP is expected to replace
diagnostic ERCP to examine the biliary and pancreatic
ducts in the near future.
Comparison with ERCP

MR cholangiopancreatography is noninvasive and
safe, because it does not require anesthesia or injection of intraductal or intravenous contrast agent. On
current MR imaging systems high-quality images can
be obtained consistently. It has been reported that
MRCP is useful in patients after incomplete or
unsuccessful ERCP [8]. In some patients, such as
those who have undergone surgery with biliary
E-mail address: richsem@med.unc.edu (R.C. Semelka).
enteric anastomosis or Billroth II, it may not be

possible to perform ERCP, so MRCP is the modality
of choice to evaluate these postsurgical patients [9].
Unlike ERCP, MRCP produces images of the ducts in
their natural state, because it does not involve distention of the ducts by injected contrast medium.
ERCP cannot evaluate extraductal structures directly,
whereas MRCP can be combined with conventional
MR imaging for the evaluation of extraductal disease,
such as tumors. ERCP has advantages over MRCP,
which include direct therapeutic interventional procedures that may be performed concurrent with
diagnostic imaging. ERCP is generally a safe procedure, but still associated with nonnegligible morbidity and mortality rates [10]. Also, technical
failures occur in up to 10% of cases because of
unsuccessful cannulation of the common bile duct
(CBD) or pancreatic duct [8,11]. In some institutions
MRCP is gradually replacing ERCP as a primary
diagnostic imaging modality to evaluate the biliary
system and pancreatic duct. This article reviews
recent progress of MRCP techniques and clinical
applications of MRCP for the evaluation of various
biliary and pancreatic diseases.
MRCP techniques
MR cholangiopancreatography uses heavily
T2-weighted images to visualize stationary or slowmoving fluids in the biliary system and pancreatic
duct with high signal intensity. For this purpose,
single-shot echo-train spin echo technique is used
most commonly. Echo-train spin echo technique uses
a single 90-degree pulse followed by multiple refo-
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 4 - 7
90
T. Motohara et al / Radiol Clin N Am 41 (2003) 8996
cusing 180-degree pulses. Extremely long echo trains

of 100 to 150 refocusing 180-degree pulses and long
effective TE are used to produce heavily T2-weighted
images. Because of the heavy T2-weighting of this
sequence, signals from the fluid in the biliary system
and pancreatic duct are hyperintense, whereas the
signal of background tissue is rendered hypointense,
enabling excellent depiction of the biliary system and
pancreatic duct. Single-shot echo-train spin echo
sequence can be performed as single-slab acquisition
or multiple thin-slice acquisition. In the single-slab
approach, a thick collimation (30 to 70 mm) single
section is acquired in an oblique coronal plane,
obtained in 2 to 3 seconds (Fig. 1). The single slab
can be acquired in various rotations to view the
biliary system and pancreatic duct from different
angles. ERCP-like images can be acquired without
maximum intensity projection (MIP) postprocessing.
This technique is useful to provide an overview of the
biliary system and pancreatic duct, but is not effective
at demonstrating small intraductal structures, such as
bile duct stones. This is because visualization of
small intraductal signal void structures is obscured
by surrounding intraductal high signal from fluid. It is
important also to acquire a multiple-slice thin-collimation sequence to examine the details of the
intraductal structures. Slice thickness of 3 to 4 mm
is needed to detect small intraductal stones. Threedimensional reconstruction may be performed by
MIP postprocessing from the thin-collimation source
images (see Fig. 1). Volume-averaging effects can
obscure small stones and subtle mural irregularity,
however, so source thin section images must always
be reviewed [1].
Usually, to obtain ERCP-like MRCP images, single-shot echo-train spin echo sequence is performed
by using a very long effective TE value (eg, 1000 milliseconds) without fat-suppression technique or a long
effective TE value (eg, 250 to 400 milliseconds) with
fat-suppression technique. With these images, however, it is not possible to evaluate periductal structures, such as tumors, which may cause narrowing or
obstruction of the ducts. Also, fluids with relatively
short TE, such as concentrated bile or mucinous
fluid, may produce very little signal with long
effective TE sequences, and that may obscure small
bile ducts or mucinous lesions. To overcome those
drawbacks of MRCP with long effective TE, an
intermediate effective TE (80 to 100 milliseconds)
can be used. This produces images where not only
all fluid including concentrated bile and mucinous
fluid is bright, but also periductal structures are well
depicted. The combination of images with intermediate effective TE and ERCP-like MRCP images gives
detailed evaluation of both intraductal and periductal structures.
Disease processes
Benign disease
Cystic diseases of the bile duct
Congenital cystic lesions of the bile duct can be
classified according to Todanis classification system
[12]: type I, choledochal cyst; type II, diverticulum of
extrahepatic ducts; type III choledochocele; type IV,
multiple segmental cysts; and type V, Carolis dis-
Fig. 1. Comparison of thick-slab MRCP and three-dimensional maximum intensity projection (MIP) reconstruction MRCP in a
patient with a pancreatic head carcinoma. (A) Coronal thick-slab single-shot echo-train spin echo MRCP. (B) Three-dimensional
MIP reconstruction obtained from thin-slice collimation source images. Dilatation of the biliary tree and pancreatic duct and
definition of the level of obstruction are demonstrated on both images. The three-dimensional MIP reconstructed image,
however, demonstrates more details of the biliary tree and the pancreatic duct. (From Bader TR, Semelka RC. Gallbladder and
biliary system. In: Abdominal-pelvic MRI. New York: Wiley Liss; 2002. p. 319 71; with permission.)
ease. MRCP can be effective and comparable with

ERCP for the evaluation of these lesions. Also, the
combination of MRCP and gadolinium-enhanced
T1-weighted images is useful to diagnose associated
findings, such as gallstone disease and cancer. MRCP
has been demonstrated to be effective in evaluating
choledochal cyst [13,14], choledochocele [15], and
Carolis disease [16].
Congenital variants of the biliary system
Anatomic variants of the cystic duct have received
much attention recently because of their higher risk of
complications during cholecystectomy. In a study
evaluating anatomic variants of the biliary tree,
MRCP could demonstrate accurately various variants,
such as a low cystic duct insertion, a medial cystic
duct insertion, a parallel course of the cystic and
hepatic ducts, and an aberrant right hepatic duct [17].
Pancreas divisum
Pancreas divisum is the most common anatomic
variant of the pancreas resulting from failure of fusion
of the duct of the embryonic dorsal pancreas and the
duct of the ventral pancreas [18]. As a result, the
pancreas has two separate ductal systems. The incidence of this anomaly is reported to be between 1.3%
and 6.7% of the general population [19]. In patients
with pancreas divisum, the minor papilla may cause
impaired pancreatic drainage and result in chronic
pancreatitis [20]. Another clinically relevant issue
with pancreas divisum is that at ERCP only the
ventral duct can be cannulated through the major
papilla, and a small ventral duct may be misdiagnosed as an obstructed pancreatic duct. On MRCP
91
images, pancreas divisum can be diagnosed by the

finding of dorsal dominant pancreatic duct running
anteriorly to the CBD and draining into the minor
papilla. A study evaluating 108 patients who underwent both ERCP and MRCP demonstrated exact
correlation between the two techniques for the depiction and exclusion of pancreas divisum [21].
Cholecystolithiasis
The primary imaging modality for cholecystolithiasis is sonography. MRCP, however, is highly
sensitive and accurate in diagnosing cholecystolithiasis and can outperform ultrasound and CT [1]. The
most reliable approach to detect gallstones with
MRCP is with the use of single-shot T2-weighted
sequences, such as MRCP (Fig. 2).
Choledocholithiasis
Accurate diagnosis of stones in the biliary ducts is
crucial because their presence is a difficult challenge
for cholecystectomy. Ultrasound and CT imaging
show relatively low sensitivity and accuracy for the
diagnosis of bile duct stones [22 25]. ERCP is
considered the gold standard procedure for the evaluation of the biliary system and has a major advantage
over other imaging modalities because of its ability to
perform therapeutic interventions and diagnosis.
Even in diagnostic ERCP alone, however, the rate
of major complications or death is not negligible and
the rate of failed ERCP is 5% to 20% [10,26,27].
MR cholangiopancreatography has been shown to
be an excellent method for detecting bile duct stones
(see Fig. 2). It is superior to CT or ultrasound and
comparable or superior to ERCP in detecting bile
Fig. 2. Gallstone disease. (A) Coronal thin-slice single-shot echo-train spin echo MRCP in a patient with gallstones and common
bile duct (CBD) stones. Multiple gallstones and CBD stones are clearly demonstrated (arrows). (B) Coronal thin-slice single-shot
echo-train spin echo MRCP in a patient with a CBD stone after liver transplantation. A CBD stone is demonstrated in the dilated
graft CBD. (From Bader TR, Semelka RC. Gallbladder and biliary system. In: Abdominal-pelvic MRI. New York: Wiley Liss;
2002. p. 319 71; with permission.)
92
duct stones [1,2,4,28]. On thin-slice source images,

stones appear as signal void lesions and can be
detected as small as 2 mm in dilated and nondilated
ducts [1]. On thick-slab images, large- or mediumsized stones in normal-caliber ducts are easily detectable, but small stones that are completely surrounded
by fluid may be obscured and difficult to detect
because of volume-averaging effects.
There are several pitfalls and mimickers of stones
with MRCP. Intraductal air bubbles (pneumobilia)
may mimic the appearance of stones. An important
differentiating feature from stones is that air bubble
filling defects lie on the nondependent portion of the
bile duct against the wall on axial images. Blood
clots may appear indistinguishable from bile duct
stones. Other pitfalls that may mimic bile duct stones
include (1) tortuosity of the bile duct running in and
out of the imaging plane; (2) merging of the cystic
duct into the CBD when observed en face on coronal
images, which may result in a round hypointense
focus; (3) metallic clips; and (4) extraductal compression from the right hepatic or the gastroduodenal
artery, which may result in a signal void focus [4,29].
Correct diagnosis usually can be achieved by careful
attention to the exact location of these foci and
interpretation of thick-slab MRCP or MIP reconstructed images in conjunction with the thin-slice
source images.
Primary sclerosing cholangitis
Primary sclerosing cholangitis is characterized by
chronic fibrosing inflammation of the biliary system
of unknown etiology. The diagnosis of primary
sclerosing cholangitis is made by cholangiographic
findings supported by histologic results. The
imaging appearance of primary sclerosing cholangitis is characterized by multiple, irregular strictures
and saccular dilatations of the intrahepatic and
extrahepatic bile ducts producing a beaded appearance. The conventional imaging modality for the
diagnosis of primary sclerosing cholangitis is ERCP.
Complications from ERCP, however, may result in
progression of cholestasis in patients with primary
sclerosing cholangitis [19,30]. MRCP has been
shown to be useful for the diagnosis and followup of primary sclerosing cholangitis (Fig. 3) [31,32].
A study evaluating MRCP in patients with primary
sclerosing cholangitis demonstrated that MRCP has
shown a sensitivity and specificity to depict primary
sclerosing cholangitis of 85% to 88% and 92% to
97%, respectively [32]. Diagnostic challenges
include that subtle changes of mild primary sclerosing cholangitis may be difficult to detect by current
MR imaging techniques, and cirrhosis may cause
Fig. 3. Primary sclerosing cholangitis. Coronal single-shot

echo-train spin echo MRCP demonstrates multiple irregular
strictures and dilatations of the biliary tree. (From Bader TR,
Semelka RC. Gallbladder and biliary system. In: Abdominalpelvic MRI. New York: Wiley Liss; 2002. p. 319 71;
with permission.)
distortion of the intrahepatic bile ducts and mimic

primary sclerosing cholangitis. A major strength is
that MRCP provides visualization of bile ducts
proximal to even severe stenoses, which may not
be evaluable by ERCP.
Postsurgical biliary complications
The most common postsurgical biliary complication is benign biliary stricture [33,34]. MRCP can
visualize the biliary tree distal and proximal to a highgrade stricture or complete obstruction. The bile ducts
distal to a stenosis, however, may be collapsed and
nonvisualized on MIP-reconstructed images leading
to overestimation of the stricture. Thin-section source
images must be used to evaluate the extent of highgrade stenoses, because even small amounts of fluid
in collapsed ducts can be depicted on these images.
Other postsurgical biliary complications include
retained bile duct stones, biliary leak, and biliary
fistula. These conditions can be evaluated effectively
by MRCP.
In patients with biliary-enteric anastomoses, it
may be difficult or impossible to perform ERCP.
On the other hand, MRCP is very effective in
evaluating the anatomy of the anastomosis, strictures
of the anastomosis, strictures of the biliary ducts, and
biliary stones proximal to the anastomosis, in up to
100% of patients [1,35]. Thin-section source images
should be examined thoroughly because the biliaryenteric anastomosis and stones may be obscured on
thick-slab and MIP-reconstructed images by the high
signal intensity of surrounding bile and bowel fluid.
Also, metallic surgical clips and pneumobilia can also
93
Neoplastic diseases
Fig. 4. Chronic pancreatitis. Coronal single-shot echo-train

spin echo thin section MRCP source image shows dilated
main pancreatic duct and its side branches. (From Bader
TR, Semelka RC. Gallbladder and biliary system. In:
Abdominal-pelvic MRI. New York: Wiley Liss; 2002.
p. 319 71; with permission.)
produce artifacts that should not be mistaken as

stones or strictures.
Chronic pancreatitis
On ERCP typical findings of chronic pancreatitis
include dilatation, narrowing or stricture, or irregularity of the pancreatic duct [30]. Prominent dilatation
of side branches is a feature of chronic pancreatitis
(Fig. 4) that helps distinguish this entity from
obstructed pancreatic duct caused by pancreatic cancer. A study evaluating 30 patients with chronic
pancreatitis undergoing ERCP and MRCP demonstrated sensitivity and specificity of 91% and 92%,
respectively, and excellent correlation between ERCP
and MRCP was reported [36].
Cholangiocarcinoma
Cholangiocarcinoma can be classified into three
types according to the anatomic location: (1) peripheral type, originating from peripheral bile ducts in
the liver; (2) hilar type (Klatskins tumor), originating from the confluence of the right and left
hepatic ducts; and (3) extrahepatic type, originating
from the main hepatic ducts, common hepatic duct,
or CBD [37,38]. Ductal obstruction is observed in
all cases of Klatskins tumor and extrahepatic cholangiocarcinoma. Evaluation of the level of obstruction is important for treatment planning. In a study
evaluating malignant perihilar biliary obstruction in
40 patients including 26 Klatskins tumors, it was
reported that MRCP was as effective as ERCP at
detecting the presence and the level of biliary
obstruction (40 of 40 cases on MRCP and 38 of
38 cases on ERCP) [39]. The disadvantage of ERCP
is that it may result in sepsis caused by overdistention of an obstructed biliary duct with stagnant bile
colonized by bacteria, and additionally ERCP may
be unable to provide sufficient biliary opacification
to evaluate adequately the region of narrowing.
MRCP, however, can demonstrate the bile duct
proximal to the obstructing site safely and efficiently
(Fig. 5).
The authors routine procedure is to evaluate
patients with possible cholangiocarcinoma with tissue
imaging sequences in addition to MRCP. Special
emphasis is made on T1-weighted fat-suppressed
spoiled gradient echo acquired 2 to 5 minutes following gadolinium administration because it is the
most consistent technique to demonstrate cholangio-
Fig. 5. Cholangiocarcinoma. (A) Coronal single-shot echo-train spin echo thin section MRCP source image shows an irregular
stricture of the common bile duct (CBD) caused by cholangiocarcinoma (arrow). (B) Transverse 2-minute postgadolinium fatsuppressed spoiled gradient echo image shows circumferential thickening and moderate enhancement of the extrahepatic CBD
caused by cholangiocarcinoma (arrow).
94
carcinoma, which appears as moderately enhancing

tissue (see Fig. 5) [40].
Pancreatic cancer
Typical pancreatographic features of pancreatic
cancer include irregular narrowing or obstruction of
the main pancreatic duct and dilatation proximal to
the lesion. Pancreatic head tumors also result in
obstruction of the CBD. MRCP is able to evaluate
the pancreatic duct proximal to an obstructing site
that ERCP may be unable to demonstrate (Fig. 6). In
a study evaluating 124 patients with a suspicion of
pancreatic cancer, MRCP was as effective as ERCP
for the detection of the pancreatic cancer with sensitivity and specificity of 84% and 97%, respectively,
for MRCP, and sensitivity and specificity of 70% and
94%, respectively, for ERCP [41]. As with other
malignant tumors, when pancreatic ductal adenocarcinoma is suspected clinically, the authors routinely
perform tissue imaging sequences. T1-weighted
spoiled gradient echo acquired immediately follow-
ing gadolinium administration is the most consistent

technique to demonstrate pancreatic cancer (see
Fig. 6) [42].
Future directions
Using current techniques, MRCP does not provide
dynamic information about pancreatic exocrine function; however, research is ongoing to obtain such
information using secretin stimulation, and results
seem promising [43]. In a study evaluating 31
patients with chronic pancreatitis and 84 patients with
suspicion of pancreatic disease, MRCP after secretin
stimulation showed reduced duodenal filling in
patients with severe chronic pancreatitis [43].
Another advance in MRCP is the use of contrast
agents that are hepatocyte-selective and eliminated, at
least in part, by the biliary system. With these agents
and faster acquisition with thin-section three-dimensional T1-weighted images of the biliary system,
Fig. 6. Pancreatic head adenocarcinoma. (A) Coronal T2-weighted single-shot echo-train spin echo image shows the dilated
common bile duct (CBD) and the pancreatic head adenocarcinoma adjacent to the ampulla of Vater. (B,C) Coronal T2-weighted
fat-suppressed thin-section MRCP shows obstruction of the CBD and main pancreatic duct with the dilatation of the ducts
proximal to the obstructed sites. (D) Transverse immediate postgadolinium spoiled gradient echo image shows low signal
intensity mass of the pancreatic head.
demonstration of smaller intrahepatic biliary branches

is feasible [44]. This approach may also facilitate
detection of functional obstruction or bile duct leak or
injury [44].
Summary
Although MRCP is still an evolving technique, it
has established itself as clinically useful and comparable with ERCP for the evaluation of various biliary
or pancreatic ductal diseases. MRCP is not only
comparable with ERCP in its diagnostic ability, but
it has the tremendous advantage of being noninvasive. Furthermore, MR imaging is useful in patients
with incomplete or failed ERCP, and in patients with
certain biliary or gastrointestinal surgical procedures
it is the imaging modality of choice. ERCP will
remain an extremely important modality because of
the great clinical importance for interventional biliary
procedures with this technique. Nonetheless, MRCP
may in the near future replace most of the diagnostic
imaging of the biliary tree, with diagnostic results
even more improved with further developments of
hardware and technique.
References
[1] Fulcher AS, Turner MA, Capps GW, Zfass AM, Baker
KM. Half-Fourier RARE MR cholangiopancreatography: experience in 300 subjects. Radiology 1998;207:
21 32.
[2] Guibaud L, Bret PM, Reinhold C, Atri M, Barkun AN.
Bile duct obstruction and choledocholithiasis: diagnosis with MR cholangiography. Radiology 1995;197:
109 15.
[3] Hirohashi S, Hirohashi R, Uchida H, et al. Pancreatitis:
evaluation with MR cholangiopancreatography in children. Radiology 1997;203:411 5.
[4] Holzknecht N, Gauger J, Sackmann M, Thoeni RF,
Schurig J, Holl J, et al. Breath-hold MR cholangiography with snapshot techniques: prospective comparison with endoscopic retrograde cholangiography.
Radiology 1998;206:657 64.
[5] Reinhold C, Taourel P, Bret PM, Cortas GA, Mehta
SN, Barkun AN, et al. Choledocholithiasis: evaluation
of MR cholangiography for diagnosis. Radiology
1998;209:435 42.
[6] Soto JA, Barish MA, Yucel EK, et al. Pancreatic duct:
MR cholangiopancreatography with a three-dimensional fast spin-echo technique. Radiology 1995;196:
459 64.
[7] Taourel P, Bret PM, Reinhold C, Barkun AN, Atri M.
MR cholangiography of anatomical variants of the biliary tree. Radiology 1996;199:521 7.
95
[8] Soto JA, Yucel EK, Barish MA, Chuttani R, Ferrucci

JT. MR cholangiopancreatography after unsuccessful
or incomplete ERCP. Radiology 1996;199:91 8.
[9] Holzknecht N, Gauger J, Stehling MK, Weinzierl M,
Reiser M. Choledocholithiasis after Billroth II surgery:
MR cholangiographic diagnosis. Eur Radiol 1997;7:
520 3.
[10] Cohen SA, Siegel JH, Kasmin FE. Complications of
diagnostic and therapeutic ERCP. Abdom Imaging
1996;21:385 94.
[11] Rieger R, Wayand W. Yield of prospective, noninvasive evaluation of the common bile duct combined
with selective ERCP/sphincterotomy in 1,930 consecutive laparoscopic cholecystectomy patients. Gastrointest Endosc 1995;42:6 12.
[12] Todani T, Watanabe Y, Narusue M, Tabuchi K, Okajima K. Congenital bile duct cysts: classification, operative procedures, and review of thirty-seven cases
including cancer arising from choledochal cyst. Am J
Surg 1977;134:263 9.
[13] Matos C, Nicaise N, Deviere J, Cassart M, Metens T,
Struyven J, et al. Choledochal cysts: comparison of
findings at MR cholangiopancreatography and endoscopic retrograde cholangiopancreatography in eight
patients. Radiology 1998;209:443 8.
[14] Miyazaki T, Yamashita Y, Tang Y, Tshuchigame T,
Takahashi M, Sera Y. Single-shot MR cholangiopancreatography of neonates, infants, and young children.
[15] De Backer AI, Vad den Abbeele K, De Schepper AM,
Van Baale A. Choledochocele: diagnosis by magnetic
resonance imaging. Abdom Imaging 2000;25:508 10.
[16] Asselah T, Ernest O, Sergent G, Lhermine C, Paris JC.
Carolis disease: a magnetic resonance cholangiopancreatography diagnosis. Am J Gastroenterol 1998;93:
109 10.
[17] Taourel P, Bret PM, Reinhold C, Barkun AN, Atri M.
Anatomic variants of the biliary tree: diagnosis with
MR cholangiopancreatography. Radiology 1996;199:
521 7.
[18] Cruikshank AH, Benbow EW. Pathology of the pancreas, 2nd edition. London: Springer; 1995.
[19] Delhaye M, Cremer M. Pancreas divisum: congenital
anatomic variant or anomaly. Contribution of endoscopic retrograde dorsal pancreatography. Gastroenterology 1985;89:951 8.
[20] Quest L, Lombard M. Pancreas divisum: opinio divisa.
Gut 2000;47:317 9.
[21] Bret PM, Reinhold C, Taourel P, Guibaud L, Atri M,
Barkun AN. Pancreas divisum: evaluation with MR
cholangiopancreatography. Radiology 1996;199:
99 103.
[22] Cronan JJ. US diagnosis of choledocholithiasis: a reappraisal. Radiology 1986;161:133 4.
[23] OConnor HJ, Hamilton I, Ellis WR, Watters J, Lintott
DJ, Axon AT. Ultrasound detection of choledocholithiasis: prospective comparison with ERCP in the postcholecystectomy patient. Gastrointest Radiol 1986;11:
161 4.
96
[24] Panasen P, Partanen K, Pikkarainen P, Alhava E,

Pirinen A, Janatuinen E. Ultrasonography, CT, and
ERCP in the diagnosis of choledochal stones. Acta
Radiol 1992;33:53 6.
[25] Stott MA, Farrand PA, Guyer PB, Dewbury KC,
Browning JJ, Sutton R. Ultrasound of the common bile
duct in patients undergoing cholecystectomy. J Clin
Ultrasound 1991;19:73 6.
[26] Duncan HD, Hodgkinson L, Deakin M, Green JR. The
safety of diagnostic and therapeutic ERCP as a daycase
procedure with a selective admission policy. Eur J Gastroenterol Hepatol 1997;9:905 8.
[27] Loperfido S, Angelini G, Benedetti G, Chilovi F,
Costan F, De Berardinis F, et al. Major early complications from diagnostic and therapeutic ERCP: a prospective multicenter study. Gastrointest Endosc 1998;
48:1 10.
[28] Regan F, Fradin J, Khazan R, Bohlmann M, Magnuson
T. Choledocholithiasis: evaluation with MR cholangiography. AJR Am J Roentgenol 1996;167:1441 5.
[29] Irie H, Honda H, Kuroiwa T, Yoshimitsu K, Aibe H,
Shinozaki K, et al. Pitfalls in MR cholangiopancreatographic interpretation. Radiographics 2001;21:23 37.
[30] Barish MA, Soto JA. MR cholangiopancreatography:
techniques and clinical applications. AJR Am J Roentgenol 1997;169:1295 303.
[31] Ernst O, Asselah T, Sergent G, et al. MR cholangiography in primary sclerosing cholangitis. AJR Am J
Roentgenol 1998;171:1027 30.
[32] Fulcher AS, Turner MA, Franklin KJ, Shiffman ML,
Sterling RK, Luketic VA, et al. Primary sclerosing
cholangitis: evaluation with MR cholangiographya
case-control study. Radiology 2000;215:71 80.
[33] Laghi A, Pavone P, Catalano C, Rossi M, Panebianco
V, Alfani D, et al. MR cholangiography of late biliary
complications after liver transplantation. AJR Am J
Roentgenol 1999;172:1541 6.
[34] Lillemoe KD, Pitt HA, Cameron JL. Current management of benign bile duct strictures. Adv Surg 1992;
25:119 73.
[35] Pavone P, Laghi A, Catalano C, Broglia L, Panebianco
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
V, Messina A, et al. MR cholangiography in the examination of patients with biliary-enteric anastomoses.

Sica GT, Braver J, Cooney MJ, Miller FH, Chai JL,
Adams DF. Comparison of endoscopic retrograde cholangiopancreatography with MR cholangiopancreatography in patients with pancreatitis. Radiology 1999;
210:605 10.
Soyer P, Bluemke DA, Reichle R, Calhoun PS, Bliss
DF, Scherrer A, et al. Imaging of intrahepatic cholangiocarcinoma: 1. Peripheral cholangiocarcinoma. AJR
Am J Roentgenol 1995;165:1427 31.
Soyer P, Bluemke DA, Reichle R, Calhoun PS, Bliss
DF, Scherrer A, et al. Imaging of intrahepatic cholangiocarcinoma: 2. Hilar cholangiocarcinoma. AJR Am J
Roentgenol 1995;165:1433 6.
Yeh TS, Jan YY, Tseng JH, Chiu CT, Chen TC, Hwang
TL, et al. Malignant perihilar biliary obstruction: magnetic resonance cholangiopancreatographic findings.
Am J Gastroenterol 2000;95:432 40.
Bader T, Semelka RC. Gallbladder and biliary sytem.
In: Abdominal-pelvic MRI. New York: Wiley-Liss;
2002. p. 319 71.
Adamek HE, Albert J, Breer H, Weitz M, Schilling D,
Riemann JF. Pancreatic cancer detection with magnetic
resonance cholangiopancreatography and endoscopic
retrograde cholangiopancreatography: a prospective
controlled study. Lancet 2000;356:190 3.
Semelka RC, Nagase LL, Armao D, Balci NC. Pancreases. In: Abdominal-pelvic MRI. New York: Wiley-Liss; 2002. p. 391.
Manfredi R, Costamagna G, Brizi MG, Maresca G,
Vecchioli A, Colagrande C, et al. Severe chronic pancreatitis versus suspected pancreatic disease: dynamic
MR cholangiopancreatography after secretin stimulation. Radiology 2000;214:849 55.
Lee VS, Rosfsky NM, Morgan GR, Teperman LW,
Krinsky GA, Berman P, et al. Volumetric mangafodipir
trisodium-enhanced cholangiography to define intrahepatic biliary anatomy. AJR Am J Roentgenol 2001;
176:906 8.
Radiol Clin N Am 41 (2003) 97 114
MR cholangiopancreatography:
evaluation of common pancreatic diseases
Laura M. Fayad, MDa,*, Thomas Kowalski, MDb, Donald G. Mitchell, MDa
a
Department of Radiology and Radiological Science, Johns Hopkins Medical Institutions, 601 North Wolfe Street,
b
Gastrointestinal Endoscopy, Division of Gastroenterology and Hepatology, 132 South 10th Street,
480 Main Building, Philadelphia, PA 19107, USA
In the initial evaluation of common pancreatic

disorders, MR cholangiopancreatography (MRCP)
has replaced the use of diagnostic endoscopic retrograde cholangiopancreatography (ERCP) at many
institutions. MRCP is a heavily T2-weighted MR
sequence that depicts the fluid-containing pancreatic
duct in a noninvasive manner, avoiding potential
complications associated with ERCP. MRCP is useful
in the setting of pancreatitis for the identification of
aberrant ductal anatomy and complications, and is
valuable for detecting and characterizing cystic pancreatic masses. Furthermore, with the addition of
conventional T1-weighted, T2-weighted, and gadolinium-enhanced sequences to MRCP, pancreatic
adenocarcinoma can also be detected and staged.
This article reviews MRCP with regard to the evaluation of common pancreatic diseases, with emphasis
on its use for guiding treatment options.
Technique
The normal pancreatic duct is a small structure,
with a diameter of 3 mm or less, and is challenging
to visualize completely by MR imaging. Because
pancreatic ductal fluid has a long T2 relaxation time,
a heavily T2-weighted sequence will result in high
signal within the pancreatic duct, whereas background tissue, which has a shorter T2 relaxation
E-mail address: lfayad1@jhmi.edu (L.M. Fayad).
time, is suppressed. The ideal sequence will provide

fast imaging, heavy T2 weighting for good duct-tobackground contrast and adequate spatial resolution
for identification of subtle ductal pathology. A number of techniques have been employed to achieve
heavy T2 weighting, including steady-state free precession gradient-recalled echo imaging [1 4], twodimensional fast spin echo (2D FSE) [5 12],
three-dimensional fast spin echo (3D FSE) [13 15],
single-shot fast spin echo (SSFSE) [16 24], and a
recently described method combining 3D FSE with
echo-planar imaging [25]. SSFSE has become the
most widely used sequence, because it provides ultrafast, reliable, MRCP imaging. Unfortunately, the
trade-off of fast sequences is a loss of spatial resolution. Although ERCP offers superior spatial resolution, the resolution of SSFSE is increased when a
512 matrix acquisition or a small field of view is
employed. In our experience, a duct with a diameter
less than 1 mm can be seen, using a field of view of
24 cm2 and a matrix size of 256 224 pixels. This
level of resolution is acceptable for the evaluation of
adults, but is insufficient for pediatric pancreaticobiliary imaging [26]. MRCP in the pediatric population requires a higher matrix or a smaller field
of view.
The SSFSE sequence can be implemented as a
series of single thick-slab acquisitions [18,19] or as a
thin multislice acquisition [18 21]. Single and multislice acquisition methods have been shown to provide
complementary data for the evaluation of the pancreas [27,28], and are superior to older techniques of
pancreatic evaluation [26].
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 6 - 0
98
L.M. Fayad et al / Radiol Clin N Am 41 (2003) 97114
Fig. 1. Thick-section MRCP. (A) Coronal, oblique, thick-section SSFSE image showing a normal pancreatic duct ( P) with a loop
configuration (L). Also shown are the common bile duct (B), common hepatic duct (H ), and the cystic duct (C ). (B) Coronal,
oblique, thick-section SSFSE image of the patient in Fig. 1, acquired at a different obliquity, showing unfolding of the pancreatic
duct loop configuration (L).
The single-slice thick-slab acquisition method is

attractive because a snapshot of the pancreaticobiliary
system is obtained while respiratory and bowel
motion is virtually eliminated, and no postprocessing
of the images is required (Fig. 1). Typically, 30-mm
to 50-mm thick slabs are prescribed in several oblique
planes to depict the extrahepatic biliary and pancreatic ducts, each requiring less than a 2-second breath
hold. Alternatively, images may be acquired at end
expiration during relaxed natural breathing. Regardless, it is important to wait at least 10 seconds
between acquisitions, to allow for recovery from the
previous overlapping radial excitation. An imaging
plane parallel to the pancreatic duct in the body and
tail of the pancreas is employed and prescribed from
an axial image (Fig. 2). Because fat is bright on
SSFSE images, chemically selective fat saturation is
utilized to increase duct-to-background contrast. A
significant advantage of the single-slice method is its
short acquisition time, which allows for the performance of a dynamic MRCP. Repeated sequential
imaging of the same slab demonstrating the pancreatic duct and extrahepatic biliary tree, performed over
several minutes, will resolve the possibility of sphincter of Oddi dysfunction. It may also demonstrate
changes in the pancreatic duct following secretin
administration [29 31].
The main drawback of the single-section acquisition is that ductal visibility may be degraded by
overlap with other fluid-containing structures or ascites included in the field of view (Fig. 3). To overcome
overlap by fluid in the stomach, ideally, patients
should fast 4 hours before the examination or be
given a T2-negative oral contrast agent such as high-
concentrate ferric ammonium citrate [32]. Negative

oral contrast may interfere with identification of the
duodenum and ampulla of Vater, however. Overlap
also can be overcome by tailoring the orientation and
positioning of the thick slab to the patients ductal
anatomy. Alternatively, a multislice thin-section
acquisition can be performed.
A multislice acquisition is typically performed
with 4-mm or 5-mm thick slices with a shorter echo
time (TE) than is used for single-section thick-slab
acquisition (see Fig. 2). With intermediate TE, (100
300 milliseconds), fluid is bright and periductal
structures are well seen, a feature particularly useful
when malignant obstruction is suspected or an overlap artifact is noted on single-section acquisition.
Fig. 2. Thin-section MRCP. Axial, thin-section SSFSE

image showing a dilated pancreatic duct (arrow). A thicksection coronal oblique image may be prescribed parallel to
the main pancreatic duct, as defined by the white rectangle.
99
Fig. 3. MRCP with overlap of fluid-containing structures. (A) Coronal, oblique, thick-section SSFSE image in a patient with
autosomal-dominant polycystic kidney disease and multiple small liver cysts (small arrows) and renal cysts (R) depicted on the
thick-section MRCP. These fluid-containing structures obscure view of the pancreatic duct ( P) in the pancreatic tail. (B) Axial,
thin-section SSFSE image of the same patient shows delineation of the pancreatic duct ( P) in the pancreatic tail. Enlarged
kidneys containing multiple cysts (R) are noted bilaterally.
Image quality is unaffected by motion because each

slice is acquired in less than 1 second, but motion will
cause slice-to-slice misregistration. Even thinner sections of the pancreaticobiliary tree (as thin as 2 mm),
can be performed and used for reconstructing a 3D
data set in any plane via maximum-intensity projection, although the quality of projection images is
usually superior with a fast thick-slab acquisition.
Secretin-enhanced dynamic MRCP (s-MRCP) is a
technique for functional imaging and improved anatomic depiction of the pancreatic ductal system.
Secretin stimulates the secretion of fluid and bicarbonate by the exocrine pancreas, with a consequent
increase in the volume of fluids inside the pancreatic
ducts [33]. In the first 5 minutes, secretin also causes
the sphincter of Oddi to contract, resulting in temporary increased pancreatic ductal pressure in healthy
subjects [34]. A thick-slab acquisition, showing the
full length of the pancreatic duct, extrahepatic biliary
tree, and duodenum in one projection, can be performed dynamically with good temporal resolution to
evaluate flow of pancreatic fluid from the pancreatic
ducts into the duodenum. Secretin improves visualization of the pancreatic duct and reduces the
false-positive depiction of strictures [29,31,35,36].
Administration of secretin also provides an estimate
of pancreatic exocrine function [31,37 39] and better
evaluation of sphincter and ductal anatomy, including
pancreas divisum. The main limitation to the performance of s-MRCP is the additional cost and limited
availability of secretin.
Evaluation of pancreatic disease:

MRCP or ERCP?
Several studies have found MRCP comparable
with ERCP for diagnosing extrahepatic biliary and
pancreatic ductal abnormalities [2,9,15,20,29,40
43]. At many institutions, MRCP has replaced ERCP
for some indications. MRCP is noninvasive, whereas
ERCP-related morbidity and mortality is not trivial,
with potential complications including pancreatitis
(3.9%), hemorrhage (1%), perforation (1%), sepsis
(0.5%), and even death in up to 0.5% of cases
[44,45]. In addition, the risk of sedation-related
complications looms. MRCP is also less expensive
than is ERCP, uses no ionizing radiation, and is less
dependent on the operators skill. No preparation is
required for MRCP (other than brief fasting at some
centers), and no exogenous contrast is needed. Furthermore, MRCP shows the ductal diameter more
accurately than does ERCP, because contrast injection during ERCP may increase biliary duct caliber
by as much as 6 mm [46,47], falsely giving the
impression of ductal dilation.
MRCP has a high success rate, whereas ERCP
failure rates range between 3% and 10% [44,48].
Failure of ERCP may be due to limited operator skill.
In the case of duodenal and gastric obstruction,
periampullary diverticula, and in patients who have
had an operative choledochoenteric or pancreaticoenteric anastomosis, however, ERCP may be technically impossible and fail in up to 20% of cases (Fig. 4)
100
lack of spatial resolution of conventional MRCP [30],

but further technical developments are needed to
adequately rival the spatial resolution of ERCP.
Pancreatic ductal anatomy and the significance of

anatomic variants
Fig. 4. Duodenal diverticulum. Coronal, oblique, thicksection SSFSE image showing a duodenal diverticulum with
an air-fluid level (long arrow) lateral to a normal common bile
duct (B). The pancreatic duct ( P) also appears normal. A
duodenal diverticulum is a common cause for ERCP failure.
[18,49]. With MRCP, a duct can be visualized beyond

an obstructing lesion. Combined with traditional
T1-weighted and T2-weighted sequences, MRCP
also allows for complementary imaging of extraductal disease. Thus, in most patients, and especially in
the 40% to 70% of patients who undergo ERCP who
have normal studies [50,51], MRCP appears to be an
excellent first choice in the workup of patients with
pancreatic disease and should be considered the test
of choice in all patients with failed or incomplete
ERCP [49,52,53].
There are two main limitations of MRCP. First,
unlike ERCP, MRCP offers no therapeutic options at
the time of diagnosis. Such options include sphincterotomy, endoscopic lithotomy, brush cytology, collection of pancreatic juice, stricture dilation, stent
placement, and biopsy. Proponents of MRCP, however, argue that MRCP provides guidance for these
therapeutic alternatives. Second, the higher level of
spatial resolution achievable by ERCP may be of
critical importance in situations in which precise
delineation of the pancreatic side branches is needed.
Such situations include the recently advocated lessinvasive pancreatic surgeries, segmental pancreatic
resection, and cyst enucleation, in which management
of the pancreatic duct and its ductules is vital to
prevent postoperative pancreatic leaks leading to
fistula formation, abscess, and hemorrhage. Usually,
more precise definition of the pancreatic side
branches can be attained by ERCP than by MRCP
[54]. The development of s-MRCP may offset the
The normal pancreatic duct, less than 3 mm in

caliber, is challenging to visualize completely by
MRCP. The caliber of the duct increases slightly
from the pancreatic tail to the head. The main duct
receives 20 to 35 short tributaries that enter perpendicularly, but are not usually seen in the normal
pancreas by MRCP. The pancreatic duct course varies
greatly, but it most commonly descends. It can have a
loop configuration, particularly at the point of fusion
of the ducts of Santorini and Wirsung in the pancreatic neck (see Fig. 1) [28].
Patterns of drainage of the pancreas also vary. In
90% of cases, the pancreas drains primarily through
the duct of Wirsung, which joins the bile duct at the
major papilla [28]. An accessory duct of Santorini
that drains through the minor papilla is present in
44% of individuals, and is not always visualized by
MRCP, due to its limited spatial resolution. The
normal main pancreatic duct, however, is visualized
in more than 80% of patients, depending on the
Fig. 5. Choledochal cyst and anomalous pancreaticobiliary

junction. Coronal, oblique, thick-section SSFSE image
showing an anomalous union of the common bile duct with
the pancreatic duct ( P) and a long common channel (long
arrow), in a patient with a choledochal cyst (C ).
MRCP sequence used [28,55]. Fulcher and Turner

[28] depicted the main pancreatic duct in the head
and body in 97% of cases and in the tail in 83% of
cases. Fortunately, an abnormally dilated duct can be
seen in 100% of cases [40].
In 1.5% to 3% of individuals, there is an anomalous union of the pancreatic and bile ducts, which
results in an unusually long common channel proximal to the duodenum. This channel is present in 33%
to 83% of patients with choledochal cysts and is
important to identify, because its presence may alter
the operative approach undertaken at surgical resection (Fig. 5) [56,57]. Gallbladder carcinoma is also
more frequent in patients with an anomalous union
than in those without such a union. Elnemr et al [58]
reported that 18.3% of patients with gallbladder
cancer had an anomalous union.
The most common anatomic abnormality of the
pancreaspancreas divisumoccurs in 5% to 14%
of the population [59,60], when the dorsal (Santorini)
and ventral (Wirsung) pancreatic ducts fail to fuse
(Fig. 6). When standard cannulation of the major
papilla is performed on ERCP, only the ventral duct is
opacified, resulting in incomplete ductography. The
termination of the ventral duct can be mistaken for
occlusion of the main duct, potentially mimicking
pancreatic cancer. An astute endoscopist will recognize this pattern, but MRCP can easily demonstrate
101
the anomaly and is an accurate method for diagnosing

pancreas divisum, because it shows a continuous
dominant dorsal pancreatic duct. The accuracy of
MRCP in the diagnosis of pancreas divisum has been
demonstrated to be 100% [41].
In a patient with pancreas divisum, the minor
papilla may provide a functional obstruction, resulting
in elevated pancreatic duct pressure that may precipitate pancreatitis, because no communication with the
major papilla exists for adequate decompression. Pancreas divisum is commonly detected incidentally in
asymptomatic patients, but it occurs more frequently
in patients who present with acute recurrent pancreatitis than in the general population [61 65].
There is no consensus regarding the appropriate
endoscopic treatment for recurrent pancreatitis associated with pancreas divisum. Papillotomy of the
minor papilla appears to yield improvement in most
cases, and placement of a transpapillary pancreatic
stent has been touted as a safe and effective endoscopic treatment [66]. A potential role of MRCP in
the diagnosis of pancreas divisum is to identify a
subset of patients with pancreas divisum and pancreatitis who may benefit from these invasive treatments. Such a subset includes patients with a true
functional obstruction at the level of the minor
papilla with or without a santorinicelea cystic
dilation of the distal dorsal duct just proximal to the
minor papilla. A santorinicele is believed to result
from relative obstruction and weakness of the distal
ductal wall and has been suggested as a possible
cause of relative stenosis of the accessory papilla
[67,68]. Manfredi et al [30] showed that a santorinicele is associated with a partial functional obstruction
at the level of the minor papilla. Following secretin
administration, the onset of duodenal filling was
delayed significantly in patients with pancreas divisum and a santorinicele compared with patients with
pancreas divisum alone. Furthermore, after sphincterotomy of the minor papilla, the size of the main
pancreatic duct and of the santorinicele was significantly reduced, and patients had symptomatic
improvement. In this study [30], conventional MRCP
detected fewer cases of pancreas divisum with or
without santorinicele compared with s-MRCP (50%
and 57%, respectively).
The role of MRCP in pancreatitis

Fig. 6. Pancreas divisum. Coronal, oblique, thick-section
SSFSE image showing pancreas divisum with the dorsal (D)
and ventral (V ) ducts. Cannulation of the major papilla (long
arrow) during ERCP results in opacification of the small
ventral duct only.
What is the state of the pancreaticobiliary tree in

patients with acute and chronic pancreatitis? This
frequent indication for ERCP may now be answered
by MRCP. Sica et al [69] showed sensitive detection
102
Fig. 7. Acute pancreatitis secondary to stones. (A) Coronal, oblique, thick-section SSFSE image showing gallstones (short
arrows) and common bile duct stones (long arrows) in the setting of acute pancreatitis. (B) Axial, thin-section SSFSE image
showing gallstones (short arrow) and common bile duct stones (long arrow). The pancreas ( P) is enlarged and heterogeneous in
this patient with acute pancreatitis.
and accurate characterization of duct segments with

MRCP that were comparable with ERCP.
In the setting of a single episode of acute pancreatitis, MRCP is focused on the noninvasive detection
or exclusion of choledocholithiasis, noting that only
30% to 52% of suspected calculi are present in
patients referred for ERCP (Fig. 7) [50]. The high
incidence of negative results using ERCP suggests
that a noninvasive test such as MRCP should be used
to screen these patients to avoid unnecessary morbidity and mortality.
In the setting of recurrent acute pancreatitis,
MRCP may be used to suggest its cause and to
detect complications of pancreatitis. Possible causes
of recurrent acute pancreatitis include choledocholithiasis, pancreatic cancer, or an anatomic abnormality such as pancreas divisum. Sphincter of Oddi
dysfunction is also a cause of recurrent pancreatitis
that, in some instances, may be suggested by

MRCP [70].
In patients with chronic pancreatitis, MRCP can
be used to support a clinical diagnosis, especially in
the early stages of chronic pancreatitis. Chronic
pancreatitis is a chronic inflammatory process that
results in pancreatic parenchymal atrophy and fibrosis. Alcoholism is the cause of at least 70% of
cases. Approximately 10% of cases are attributed to
chronic ductal obstruction, autoimmune disorders,
inflammatory disease, and inherited diseases causing
abnormal pancreatic enzymes or ductal secretion.
Remaining cases are idiopathic. Alcoholic pancreatitis is usually heterogeneous and characterized by
side-branch dilation and ductal calcifications (Fig. 8),
whereas obstructive pancreatitis is more homogeneous, lacks calcifications, and is associated more
often with main duct dilation. Nonalcoholic duct-
Fig. 8. Chronic pancreatitis. (A) Coronal, oblique, thick-section SSFSE image showing a markedly dilated pancreatic duct (large
arrow) with markedly dilated side-branches (small arrows) in a patient with chronic alcoholic pancreatitis. (B) Axial, thinsection, fat-suppressed, 2D FSE image showing small hypointense filling defects in the pancreatic duct and its side branches,
representing pancreatic duct stones (arrows) in a patient with chronic alcoholic calcific pancreatitis.
destructive pancreatitisor autoimmune pancreatitisis characterized by a narrow pancreatic duct

and diffuse parenchymal abnormality, typically involving the pancreatic body and tail, without ductal calcifications [71,72].
Early alcoholic chronic pancreatitis manifests as
irregularities and dilation of the ductal side branches.
These side branches may be below the limits of
resolution of MRCP, so ERCP is generally more
sensitive to early side-branch changes. Side-branch
ectasia is the most specific and prominent feature of
alcoholic chronic pancreatitis. At a later stage, the
main pancreatic duct is dilated with loss of the normal
tapering of the duct in the tail. Areas of focal
narrowing produce a characteristic beaded chain of
lakes appearance. Even the biliary tract may become
103
dilated as a result of fibrosis in the head of the

pancreas. MRCP has demonstrated the pancreatic
duct, after stimulation with secretin, loses distensibility and has decreased exocrine function [31,37].
In addition, MRCP may be used in conjunction
with other MR sequences, especially nonenhanced
T1-weighted images, which show low pancreatic
signal intensity in patients with chronic pancreatitis.
MRCP is not only important for detecting chronic
pancreatitis, but is also valuable for the identification
of a surgically or endoscopically correctable lesion
(Fig. 9). The location of strictures, degree of ductal
dilation, presence of ductal filling defects, and associated complications such as pseudocysts all influence the therapy of patients with chronic pancreatitis.
MRCP agrees with ERCP in 83% to 100% of ductal
Fig. 9. Chronic pancreatitis. (A) Coronal, oblique, thick-section SSFSE image showing changes of chronic pancreatitis, including
a dilated pancreatic duct (large arrow) with filling defects representing mucus (small arrows), and two pseudocysts (C ). (B)
Coronal, thin-section SSFSE image showing the findings in Fig. 9A in greater detail. (C) Coronal, thin-section SSFSE image
acquired more anteriorly. C, pseudocysts. (D) Axial, nonenhanced, T1-weighted, gradient echo image showing decreased signal
in the pancreas ( P) from fibrosis. The dilated pancreatic duct (arrow) and two pseudocysts (C ) are again noted.
104
Fig. 10. Focal obstructive pancreatitis due to adenocarcinoma of the pancreas. (A) Coronal, thick-section SSFSE image showing
a dilated pancreatic duct (thick arrows) in the body and tail of the pancreas with termination (thin arrow) of the duct in the body.
(B) Axial, T1-weighted, 3D, gradient echo image of the pancreas obtained in the arterial phase following the administration of
intravenous gadolinium shows a hypoenhancing mass (arrow) that is responsible for obstruction of the pancreatic duct with
resultant obstructive pancreatitis distally.
dilation cases, in 70% to 92% of ductal narrowing

cases, and in 92% to 100% of filling defect cases
[36,43]. On MRCP, filling defectsrepresenting
mucus, calculi and debriscan be reliably identified
with a diameter as small as 2 mm [40]. MRCP is,
however, superior for detecting pseudocysts, which
are missed approximately 50% of the time by ERCP
[73], although ERCP can consistently determine the
presence or site of communication of a pseudocyst
with the main pancreatic duct.
Therapeutic options for chronic pancreatitis
include surgical decompression, partial pancreatectomy, total pancreatectomy, and endoscopic decompression, which relieve pain in 75% to 90% of
patients. Ductal decompression is the main principle
of therapy, because early decompression delays the
onset of exocrine dysfunction, as well as endocrine
dysfunction that occurs in 33% of patients with
chronic pancreatitis [74]. Filling defects such as
calculi, mucus, and debris may be removed endoscopically through a pancreatic duct sphincterotomy,
and strictures can be dilated with short-term stent
placement to improve pain [75]. Duct decompression
by surgery depends, in part, on the size of the main
pancreatic duct. For example, duct-destructive chronic pancreatitis, with a main duct diameter of less than
3 mm, requires a drainage procedure different from
that required for a duct diameter over 7 mm [75,76].
Hence, when using MRCP it is important to fully
describe the state of the pancreatic duct in the
setting of chronic pancreatitis to adequately guide

treatment options.
Chronic pancreatitis or carcinoma?

When findings of chronic pancreatitis are identified in a patient without a prior history of chronic
pancreatitis or of ethanol abuse, an obstructing lesion
should be suspected (Fig. 10). Pancreatic ductal
adenocarcinoma is the usual cause of chronic
obstructive pancreatitis and comprises 75% to 90%
of all pancreatic carcinomas [77]. Differentiating
adenocarcinoma from mass-forming chronic pancreatitis with MR imaging is sometimes difficult. Typically, the chronically inflamed pancreas will enhance
more than will pancreatic tumors on immediate postgadolinium images, particularly those tumors arising
in the head. Unfortunately, the degree of enhancement cannot be used to reliably distinguish these
entities because abundant fibrosis is seen in both
chronic pancreatitis and carcinoma, accounting for
their similar appearances [78].
MRCP may be helpful to aid in this differentiation, because chronic alcoholic pancreatitis, compared with chronic obstructive pancreatitis due to
adenocarcinoma, is more frequently associated with
an irregularly dilated duct with intraductal calcification [79]. The ratio of duct caliber to pancreatic gland
width is higher in patients with carcinoma [80]. Also,
105
Fig. 11. Pancreatic adenocarcinoma. (A) Coronal, thick-section SSFSE image showing the classic double-duct sign of
pancreatic carcinoma; both the pancreatic duct ( P) and common bile duct (C ) are dilated and abruptly terminate (large arrow) in
the head of the pancreas. In this case, the intrahepatic biliary ducts are also dilated. (B) Axial, nonenhanced, T1-weighted,
gradient echo image showing an ill-defined hypointense mass (arrow) in the head of the pancreas. A biopsy of this mass revealed
pancreatic adenocarcinoma.
the duct-penetrating sign, seen in 85% of chronic

pancreatitis and in only 4% of patients with cancer,
helps to distinguish an inflammatory pancreatic mass
from pancreatic carcinoma. The duct-penetrating
sign refers to a nonobstructed main pancreatic duct
penetrating an inflammatory pancreatic mass, unlike
its usual obstruction by pancreatic carcinoma. [81].
Furthermore, MRCP can depict the classic double
duct sign of pancreatic carcinoma (enlargement and
noncommunication of the pancreatic and common
bile ducts) and the imaging counterpart of Courvoisiers sign (an enlarged, nontender gallbladder
caused by an obstructing tumor) (Fig. 11) [82]. A
normal-sized pancreatic duct is present in up to 20%
of patients with adenocarcinoma, however, and
should not dissuade its diagnosis in the setting of
common bile duct dilation.
These latter signs are useful when present, but
MRCP (like ERCP) is thought to be a poor way to
differentiate benign from malignant strictures.
Because morphologic features of benign and malignant strictures overlap, ERCP may be the imaging
modality of choice because of its ability to obtain a
diagnostic sample with brush cytologic biopsy
[75,83]; however, MRCP, including MR imaging
pulse sequences, has a sensitivity of 84% for diagnosing pancreatic carcinoma, whereas the corresponding sensitivity for ERCP with brush cytology
varies between 33% and 85% [42,83]. Adding MRCP
to conventional T1-weighted and T2-weighted
sequences improves specificity by depicting extra-
ductal structures not seen with ERCP [84]. Comprehensive MR imaging is also useful to accurately
determine resectability [85].
Cystic pancreatic masses

The incidence of detected cystic pancreatic
masses is increasing because of the widespread use
of cross-sectional imaging. Cystic pancreatic lesions
include benign entities such as pseudocysts and
epithelial pancreatic cysts, as well as malignant
lesions. Epithelial cysts are usually associated with
entities such as polycystic kidney disease and von
Hippel-Lindau disease. Other cystic lesions are discussed below.
Pancreatic pseudocysts occur as a complication of
acute or chronic pancreatitis and represent 90% of
cystic pancreatic masses [86]. They are encapsulated
collections of pancreatic fluid, caused by pancreatic
duct disruption and tissue dissolution in acute pancreatitis, and microperforation of the pancreatic duct in
chronic pancreatitis. The surgical definition of a
pseudocyst requires that it be present for at least
6 weeks. These lesions may communicate with the
main pancreatic duct and may be identified on ERCP
(Fig. 12). Less than 50% of pseudocysts are detected
at ERCP [73], however, giving MR imaging a large
advantage for their diagnosis. Pseudocysts are usually
accompanied by a clinical history of pancreatitis and
are associated with pancreatic parenchymal and ductal
106
Fig. 12. Communicating pseudocyst. Axial, thin-section SSFSE image showing an uncomplicated pseudocyst (C) with
communication to the pancreatic duct (arrow) in a patient with pancreatitis.
changes that suggest pancreatitis, making their differentiation from pancreatic neoplasms possible in most
cases. Pseudocysts resolve spontaneously in 60% of
cases [87].
Treatment options for persistent pseudocysts
include endoscopic, radiologic, and surgical drainage.
These must be considered cautiously, in the event that
a cystic neoplasm is misdiagnosed as a pseudocyst
[88]. Misdiagnosis, usually by CT, has been reported
as high as one third of the time [89 91]. The traditional approach for treating pseudocysts that require
drainage has been surgical. Treatment is considered
when the patient is symptomatic, if the pseudocyst
demonstrates enlargement or complications including

hemorrhage, or if there is suspicion of a malignancy
[88]. Enlargement and hemorrhage are two factors that
can be determined with MR imaging (Fig. 13).
Cystic neoplasms of the pancreas are uncommon,
representing 10% of cystic lesions [86]. Classification
of cystic neoplasms is based on the location of the
lesion, the size of the cysts, the serous or mucinous
nature of the contents, and the most dedifferentiated
epithelial change recognizable at pathology [92,93].
Additionally, almost any pancreatic neoplasm can
present as a cystic mass, including adenocarcinoma,
which is the most common pancreatic neoplasm.
Fig. 13. Hemorrhage pseudocyst. (A) Axial, thin-section SSFSE image showing a large complex cystic collection with a fluid
fluid level (arrow) representing a hemorrhagic pseudocyst in a patient with pancreatitis. The head of the pancreas ( P) is denoted.
(B) Axial, T1-weighted, gradient echo image showing hyperintensity (arrows) at the posterior aspect of the pseudocyst,
representing hemorrhage. The head of the pancreas (P) is again denoted.
Parenchymal cystic lesions include serous and mucinous cystic neoplasms. Intraductal neoplasms are
referred to as intraductal papillary mucinous tumors
(IPMTs). Cystic neoplasms can easily be detected on
MRCP because of their high fluid content, but full
examination requires T1-weighted, T2-weighted, and
postgadolinium sequences.
Serous microcystic adenomas are benign pancreatic parenchymal lesions with a relatively equal
distribution throughout the pancreas. Although typically appearing solid on CT or US, a serous adenoma
is cystic with more than six internal cysts, each
measuring less than 2 cm in diameter. Approximately
40% of these tumors have calcifications and 15%
have a central stellate scar (Fig. 14). Their soft tissue
component is typically hypervascular and aspirated
contents contain glycogen [94]. A small serous
tumor adjacent to the main pancreatic duct or a
branch duct may be difficult to distinguish from an
intraductal neoplasm.
Mucinous cystic neoplasms are also parenchymal
lesions. In the past, this neoplasm was subcategorized
107
into macrocystic adenomas and adenocarcinomas,

which are indistinguishable on the basis of imaging.
In fact, all mucinous cystic neoplasms should be
considered malignant or potentially malignant; thus,
this subcategorization is not appropriate. Mucinous
neoplasms, with strict histologic criteria, probably
occur only in women and are usually located in the
pancreatic body and tail. Most mucinous cystic neoplasms have fewer than six cysts, each greater than
2 cm in diameter. Twenty-five percent of these lesions
have calcification, and the soft tissue component is
hypovascular. Aspiration of these lesions yields
mucin [92]. Because these lesions are frequently
unilocular, they may be confused with pseudocysts.
In such cases, changes of pancreatitis should be
sought to confirm the possibility of a pseudocyst
(Fig. 15).
Intraductal tumors, previously described in the
literature under different names such as ductectatic
mucinous cystadenocarcinomas, predominate in men
and older individuals. These tumors, now referred to
as IPMTs, usually grow slowly, with a good prog-
Fig. 14. Serous tumor. (A) Coronal, thick-section SSFSE image showing a round, well-defined, cystic mass (large arrow) in the
head of the pancreas with a conglomerate of small cysts measuring less than 2 cm each. The common bile duct (B), pancreatic
duct ( P), and cystic duct (C) are denoted on this image. (B) Axial, thin-section SSFSE image showing the cystic mass (thick
arrow) with a suggestion of a central scar (thin arrow), characteristic of a serous tumor.
108
Fig. 15. Mucinous tumor. (A) Coronal, thick-section SSFSE image depicting a cystic mass (thick arrow) in the tail of the
pancreas, without associated dilation of the pancreatic duct (thin arrow). (B) Axial, thin-section SSFSE image showing the cystic
mass (thick arrow) with internal complex signal. The pancreatic duct (thin arrows) is not dilated. Resection of this mass yielded a
benign mucinous cystic tumor.
nosis. IPMTs arise from the epithelial lining of the

pancreatic ductal system and include lesions representing the histologic progression of epithelial hyperplasia, dysplasia, adenoma, carcinoma-in-situ, and
invasive carcinoma [95 98]. Hyperplasia, dysplasia,
and adenoma may undergo malignant transformation
over many years [98]. Multiple IPMTs can be present
in an individualapproximately 23% of the timeas
described in a series by Megibow et al [99]. IPMT is

associated with excessive mucin secretion, resulting
in progressive ductal dilation or cyst formation
(Fig. 16). IPMT may involve the main duct or branch
ducts of the pancreas. Imaging patterns include segmental or diffuse involvement of the main pancreatic
duct, and microcystic or macrocystic masslike lesions
involving the branch ducts [95,97,98].
Fig. 16. IPMT. (A) Coronal, thick-section SSFSE image showing a cystic mass (large arrow) associated with a dilated pancreatic
duct ( P) with dilated side branches. Incidental note is made of a cystic duct remnant (C ). (B) Coronal, thin-section SSFSE image
showing the cystic mass (large arrow) in communication with a dilated draining pancreatic duct ( P), a finding that is highly
suggestive of an intraductal papillary mucinous tumor.
When an IPMT involves the full length of the

main pancreatic duct without a localized cystic mass,
differentiation from chronic pancreatitis may be difficult [100,101]. A finding virtually pathognomonic
of IPMTs is dilation of the major papilla, minor
papilla, or both, with bulging into the duodenal
lumen. This finding, which can be seen on MR
imaging and is well appreciated by ERCP, is demonstrated with CT imaging approximately 25% of the
time (Fig. 17) [97].
With segmental involvement of the main pancreatic duct, the adjacent pancreatic parenchyma is
normal or thin. IPMTs such as this can be difficult
109
to differentiate from localized chronic obstructive

pancreatitis. In such cases, hypointense filling defects
representing mucin facilitate the diagnosis of IPMT.
When the IPMT is located in the head, it may be
difficult to differentiate between diffuse invasion of
the main pancreatic duct and simple ductal dilation
from mechanical obstruction. In difficult cases, ERCP
is valuable to demonstrate intraluminal mucinous
filling defects with jellylike mucin leaking from the
papilla, another pathognomonic finding for IPMT. On
occasion, mucin can be viscous enough to obstruct
the pancreatic duct, preventing successful ERCP.
Nevertheless, the endoscopist can confirm the pres-
Fig. 17. IPMT. (A) Coronal, thick-section SSFSE image showing an IPMT (thick arrow) with a dilated draining pancreatic duct
( P) and bulging papilla (thin arrow). The latter finding is pathognomonic of IPMTs. (B) Coronal, thick-section SSFSE image
obtained in a different projection showing the dilated main pancreatic duct (thin arrows) in association with this IPMT (thick
arrow). (C) Coronal, thin-section SSFSE image again demonstrating a cystic mass representing an IPMT (thick arrow) with a
dilated draining pancreatic duct and bulging papilla (thin arrow). On thin-section MRCP images, a bulging papilla is seen as a
filling defect in the duodenum. In this case, possible papillary projections are noted within the cystic mass.
110
Fig. 18. Multiple IPMTs. Axial, thin-section SSFSE image

showing multiple cystic lesions in the pancreas, representing
multiple IPMTs (thick arrows). It is difficult to differentiate
these cystic masses from pseudocysts and other pancreatic
cystic lesions. The pancreas is atrophied, which is likely
related to chronic obstruction by mucus. The pancreatic duct
(thin arrow) is slightly prominent.
ence of an IPMT by observing copious drainage of

mucin from the papilla [97,102].
Branch-duct IPMT is most frequently located in
the uncinate process and can have a macrocystic or
microcystic appearance [97,98]. Communication with
the main pancreatic duct is a valuable finding that is
often demonstrated best on ERCP [103] and s-MRCP.
A branch-duct IPMT may be differentiated from a
communicating pseudocyst if the IPMT contains
papillary proliferations. If an IPMT manifests as a
more cystic masslike lesion, it can resemble a mucinous cystic tumor or necrotic adenocarcinoma
(Fig. 18). With the latter entities, however, the main
pancreatic duct central to the tumor should not be
dilated as it is in the presence of an IPMT.
A number of researchers have concluded that
MRCP is more sensitive and effective than is ERCP
in evaluating IPMT [54,104 107]. MRCP, however,
does not offer definitive pathologic information to
decide whether a lesion is malignant, but features
have been described that suggest malignancy. Generally, a less-favorable histology is noted with the
main-duct type of IPMT [99,108,109]. The observation of thick walls and mural nodules aids the
diagnosis of malignancy [104,108,110]; the detection
of nodules in the cystic lesion is better accomplished
with MRCP than with ERCP [104,105]. The size of
the lesion is also important: in one series, Obara et al
[111] found that 83% of tumors larger than 4 cm were
malignant. The size of the main pancreatic duct is
valuablemain pancreatic ductal dilation greater

than 15 mm [109], as well as diffuse main pancreatic
duct dilation with the branch-duct type of IPMT [95],
is associated with malignancy.
The features described above are influential in
deciding whether an IPMT is benign or malignant,
but at this time, imaging cannot reliably distinguish
benign from malignant tumors [111 113]. Surgical
management is usually recommended when these
lesions are encountered [99,114,115]. A review of
cystic pancreatic masses by Megibow et al [99],
however, concluded that surveillance might be possible if lesions are smaller than 2.5 cm, spare the main
pancreatic duct, and demonstrate no solid components. Because many of these patients are asymptomatic elderly individuals and growth may be slow or
negligible over several years, surgical removal may
not be the only appropriate management.
Summary
In the evaluation of common pancreatic diseases,
MRCP is a noninvasive alternative to ERCP. Ductal
anatomy can be ascertained without risk of complications. MRCP is valuable in defining common
anatomic variants, determining the state of the pancreatic duct in pancreatitis, and characterizing neoplasms, especially combined with other MR imaging
sequences. With the advent of MRCP, techniques
requiring endoscopy and percutaneous access are
largely reserved for histologic diagnosis and treatment, or for cases in which MRCP fails to establish
a diagnosis.
References
[1] Wallner BK, Schumacher KA, Weidenmaier W, Friedrich JM. Dilated biliary tract: evaluation with MR
cholangiography with a heavily T2-weighted contrast-enhanced fast-sequence. Radiology 1991;181:
805 8.
[2] Morimoto K, Shimoi M, Shirakawa T, et al. Biliary
obstruction: evaluation with three-dimensional MR
cholangiography. Radiology 1992;183:578 80.
[3] Ishizaki Y, Wakayama T, Okada Y, Kobayashi T.
Magnetic resonance cholangiography for evaluation
of obstructive jaundice. Am J Gastroenterol 1993;12:
2072 7.
[4] Hall-Craggs M, Allen C, Owens C, et al. MR cholangiography: Clinical evaluation in 40 cases. Radiology 1993;189:423 7.
[5] Meakem TJ, Holland GA, McDermott VG, et al. Fast
spin-echo multicoil magnetic resonance cholangiog-
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
raphy: initial experience [abstract]. In: Proceedings of

the 12th Annual Scientific Meeting of the Society of
Magnetic Resonance in Medicine. Berkeley, California: Society of Magnetic Resonance in Medicine;
1993. p. 41.
Outwater EK. MR cholangiography with a fast spinecho sequence [abstract]. J Magn Reson Imaging
1993;3(P):131.
Takehara Y, Ichijo K, Tooyama N, et al. Breath-hold
MR cholangiopancreatography with a long-echo-train
fast spin-echo sequence and a surface coil in chronic
pancreatitis. Radiology 1994;192:74 8.
Guibaud L, Bret PM, Reinhold C, Atri M, Brkum AN.
Diagnosis of choledocholithiasis: value of MR cholangiography. Am J Roentgenol 1994;163:847 50.
Guibaud L, Bret PM, Reinhold C, Atri M, Barkum AN.
Bile duct obstruction and choledocholithiasis: diagnosis with MR cholangiography. Radiology 1995;197:
109 15.
Macaulay SE, Schulte SJ, Sekijima JH, et al. Evaluation of a non-breath-hold MR cholangiographic technique. Radiology 1995;196:227 32.
Becker CD, Grossholz M, Becker M, Mentha G, de
Peyer R, Terrier F. Choledocholithiasis and bile duct
stenosis: accuracy of MR cholangiopancreatography.
Radiology 1997;205:523 30.
Reinhold C, Taourel P, Bret PM, et al. Choledocholithiasis: evaluation of MR cholangiography for diagnosis. Radiology 1998;209:435 42.
Barish MA, Yucel EK, Soto JA, Chuttani R, Ferrucci
JT. MR cholangiopancreatography: efficacy of threedimensional turbo spin-echo technique. Am J Roentgenol 1995;165:295 300.
Soto JA, Barish MA, Yucel EK, Siegenberg D, Ferrucci JT, Chuttani R. Magnetic resonance cholangiography: comparison to endoscopic retrograde
cholangiopancreatography. Gastroenterology 1996;
110:589 97.
Chan Y, Chan ACW, Lam WWM, et al. Choledocholithiasis: comparison of MR cholangiography and endoscopic retrograde cholangiography. Radiology
1996;200:85 9.
Laubenberger J, Buchert M, Schneider B, Blum U,
Hennig J, Langer M. Breath-hold projection magnetic
resonance cholangiopancreatography (MRCP): a new
method for the examination of the bile and pancreatic
ducts. Magn Reson Med 1995;33:18 23.
Kim TK, Han JK, Kim SJ, Bae SH, Choi BI. MR
cholangiopancreatography: comparison between halfFourier acquisition single-shot turbo spin-echo and
two-dimensional turbo spin-echo pulse sequences.
Abdom Imaging 1998;23:398 403.
Reuther G, Kiefer B, Tuchmann A. Cholangiography
before biliary surgery: single-shot MR cholangiography versus intravenous cholangiography. Radiology
1996;198:561 6.
Miyazaki T, Yamashita Y, Tsuchigame T, Yamamoto
H, Urata J, Takahashi M. MR cholangiopancreatography using HASTE (half-fourier acquisition single-
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
111
shot turbo spin-echo) sequences. Am J Roentgenol

1996;166:1297 303.
Regan F, Fradin J, Khazan R, Bohlman M, Magnuson
T. Choledocholithiasis: evaluation with MR cholangiography. Am J Roentgenol 1996;167:1441 5.
Holzknecht N, Gauger J, Sacmann M, et al. Breathhold MR cholangiography with snapshot techniques:
prospective comparison with endoscopic retrograde
cholangiography. Radiology 1998;206:657 64.
Irie H, Honda H, Tajima T, et al. Optimal MR cholangiopancreatographic sequence and its clinical application. Radiology 1998;206:379 87.
Yamashita Y, Abe Y, Tang Y, Urata J, Sumi S,
Takahashi M. In vitro and clinical studies of image
acquisition in breath-hold MR cholangiopancreatography: single-shot projection technique versus multislice technique. Am J Roentgenol 1997;168(6):
1449 54.
Soto JA, Barish MA, Alvarez O, Medina S. Detection
of choledocholithiasis with MR cholangiography:
comparison of three-dimensional fast spin-echo and
single and multisection half-fourier rapid acquisition
with relaxation enhancement sequence. Radiology
2000;215:737 45.
Wielopolski PA, Gaa J, Wielopolski DR, Oudkerk M.
Breath-hold MR cholangiopancreatography with
three-dimensional, segmented, echo-planar imaging
and volume rendering. Radiology 1999;210:247 52.
Irie H, Honda H, Jimi M, et al. Value of MR cholangiopancreatography in evaluating choledochal
cysts. Am J Roentgenol 1998;171:1381 5.
Masui T, Takehara Y, Ichijo K, Naito M, Watahiki H,
Kaneko M, et al. Evaluation of the pancreas: a comparison of single thick-slice MR cholangiopancreatography with multiple thin-slice volume
reconstruction MR cholangiopancreatography. Am J
Roentgenol 1999;173:1519 26.
Fulcher AS, Turner MA. MR pancreatography: a useful tool for evaluating pancreatic disorders. Radiographics 1999;19:5 24.
Matos C, Metens T, Deviere J, et al. Pancreatic duct:
morphologic and functional evaluation with dynamic
MR pancreatography after secretin stimulation. Radiology 1997;203:435 41.
Manfredi R, Costamagna G, Brizi MG, et al. Pancreas
divisum and santorinicele: diagnosis with dynamic
MR cholangiopancreatography with secretin stimulation. Radiology 2000;217:403 8.
Manfredi R, Costamagna G, Brizi M, et al. Severe
chronic pancreatitis versus suspected pancreatic disease: dynamic MR cholangiopancreatography after
secretin stimulation. Radiology 2000;214:849 55.
Takahara T, Yoshikawa T, Saeki M, et al. High concentration ferric ammonium citrate (FAC) solution as
a negative bowel contrast agent. Nippon Igaku Hoshasen Gakkai Zasshi 1995;55:425 6.
Dreiling DA, Messer J. The secretin story. Am J Gastroenterol 1978;70:455 79.
Geenen JE, Hogan WJ, Dodds WJ, et al. Intraluminal
112
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]

pressure recording from the human sphincter of Oddi.
Gastroenterology 1980;78:317 24.
Takehara Y. Can MRCP replace ERCP? J Magn Reson Imaging 1998;8:517 34.
Nicaise N, Pellet O, Metens T, et al. Magnetic resonance cholangiopancreatography: interest of IV secretin administration in the evaluation of pancreatic
ducts. Eur Radiol 1998;8:16 22.
Cappeliez O, Delhaye M, Deviere J, et al. Chronic
pancreatitis: evaluation of pancreatic exocrine function with MR pancreatography after secretin stimulation. Radiology 2000;215:358 64.
Heverhagen JT, Muller D, Battmann A, et al. MR
hydrometry to assess exocrine function of the pancreas: initial results of noninvasive quantification of
secretion. Radiology 2001;218(1):61 7.
Sho M, Nakajima Y, Kanehiro H, et al. A new
evaluation of pancreatic function after pancreaticoduodenectomy using secretin magnetic resonance
cholangiopancreatography. Am J Surg 1998;176(3):
279 82.
Fulcher AS, Turner MA, Capps GW, Zfass AM,
Baker KM. Half-fourier RARE MR cholangiopancreatography: experience in 300 subjects. Radiology
1998;207:21 32.
Bret PM, Reinhold C, Taourel P, Guibaud L, Atri
M, Barkun AN. Pancreas divisum: evaluation with
MR cholangiopancreatography. Radiology 1996;
199:99 103.
Adamek HE, Albert J, Breer H, et al. Pancreatic cancer detection with magnetic resonance cholangiopancreatography and endoscopic retrograde
cholangiopancreatography: a prospective controlled
study. Lancet 2000;356:190 3.
Soto JA, Barish MA, Yucel EK, et al. Pancreatic duct:
MR cholangiopancreatography with a three-dimensional fast spin-echo technique. Radiology 1995;
196:459 64.
Bilbao MK, Dotter CT, Lee TG, et al. Complications
of endoscopic retrograde cholangiopancreatography
(ERCP). A study of 10,000 cases. Gastroenterology
1976;70:314 20.
Halme L, Doepel M, von Numers H, et al. Complications of diagnostic and therapeutic ERCP. Ann Chir
Gynaecol 1999;88:127 31.
Cooperberg PL, Li D, Wong P, Cohen MM, Burhenne
HJ. Accuracy of common hepatic duct size in the
evaluation of extrahepatic biliary obstruction. Radiology 1980;135:141 4.
Chang VH, Cunningham JJ, Fromkes JJ. Sonographic
measurement of the extrahepatic bile duct before and
after retrograde cholangiography. Am J Roentgenol
1985;144:753 5.
Silvis S, Ansel JH. Endoscopic retrograde cholangiography: application in biliary tract disease. In: Berk JE,
editor. Bockus gastroenterology. 4th edition, vol. 6.
Philadelphia: WB Saunders; 1985. p. 3569 80.
Adamek H, Weitz M, Breer H, Jakobs R, Schilling D,
Riemann J. Value of magnetic-resonance cholangio-
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
pancreatography (MRCP) after unsuccessful endoscopic retrograde cholangiopancreatography (ERCP).

Endoscopy 1997;29:741 4.
Paul A, Millat B, Holthausen U, et al. Diagnosis and
treatment of common bile duct stones (CBDS): results
of a consensus development conference. Surg Endosc
1998;12:856 64.
Liu TH, Consorti ET, Kawashima A. The efficacy of
magnetic resonance cholangiography for the evaluation of patients with suspected choledocholithiasis
before laparoscopic cholecystectomy. Am J Surg
1999;178:480 4.
Varghese JC, Farrell MA, Courtney G, Osborne H,
Murray FE, Lee MJ. Role of MR cholangiopancreatography in patients with failed or inadequate ERCP.
Am J Roentgenol 1999;173:1527 33.
Soto JA, Yucel EK, Barish MA, Chuttani R, Ferrucci
JT. MR cholangiopancreatography after unsuccessful
or incomplete ERCP. Radiology 1996;199:91 8.
Yamaguchi K, Chijiiwa K, Shimizu S, et al. Comparison of endoscopic retrograde and magnetic resonance cholangiopancreatography in the surgical
diagnosis of pancreatic disease. Am J Surg 1998;
175(3):203 8.
Barish MA, Yucel EK, Ferrucci JT. Current concepts:
magnetic resonance cholangiopancreatography.
N Engl J Med 1999;341(4):258 64.
Sugiyama M, Baba M, Atomi Y, Hanaoka H, Mizutani Y, Hachiya J. Diagnosis of anomalous pancreaticobiliary junction: value of magnetic resonance
cholangiopancreatography. Surgery 1998;123(4):
391 7.
Mirsa SP, Dwivedi M. Pancreaticobiliary ductal
union. Gut 1990;31:1144 9.
Elnemr A, Ohta T, Kayahara M, et al. Anomalous
pancreaticobiliary ductal junction without bile duct
dilatation in gallbladder cancer. Hepatogastroenterology 2001;48(38):382 6.
Stern CD. A historical perspective on the discovery of
the accessory duct of the pancreas, the ampulla of
Vater and pancreas divisum. Gut 1986;27:203 12.
Satterfield ST, McCarthy JH, Greenen JE, et al. Clinical experience in 82 patients with pancreas divisum:
preliminary results of manometry and endoscopic
therapy. Pancreas 1988;3:248 53.
Bernard JP, Sahel J, Giovanni M, Sarles H. Pancreas
divisum is a probable cause of acute pancreatitis:
a report of 137 cases. Pancreas 1990;5:248 54.
Cotton PB. Congenital anomaly of pancreas divisum
as a cause of obstructive pain and pancreatitis. Gut
1980;21:105 14.
Gregg JA. Pancreas divisum: its association with pancreatitis. Am J Surg 1977;134:539 43.
Startitz M, Meyer zum Buschenfelde KH. Elevated
pressure in the dorsal part of pancreas divisum: the
cause of chronic pancreatitis. Pancreas 1988;3:
108 10.
Warshaw AL, Simeone J, Schapiro RH, Hedberg SE,
Mueller PE, Ferrucci JU. Objective evaluation of am-
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
pullary stenosis with ultrasonography and pancreatic

stimulation. Am J Surg 1985;149:65 71.
Ertan A. Long-term results after endoscopic pancreatic stent placement without pancreatic papillotomy in
acute recurrent pancreatitis due to pancreas divisum.
Gastrointest Endosc 2000;52(1):9 14.
Siebert DG, Matulis SR. Santorinicele as a cause of
chronic pancreatic pain. Am J Gastroenterol 1995;
90:121 3.
Eisen G, Schutz S, Metzler D, Baillie J, Cotton PB.
Santorinicele: new evidence for obstruction in pancreas divisum. Gastrointest Endosc 1994;40:73 6.
Sica GT, Braver J, Cooney MJ, Miller FH, Chai JL,
Adams DF. Comparison of endoscopic retrograde
cholangiopancreatography with MR cholangiopancreatography in patients with pancreatitis. Radiology
1999;210:605 10.
Pamos S, Rivera P, Canelles P, et al. Magnetic resonance cholangiopancreatography (MRCP) versus endoscopic retrograde cholangiopancreatography
(ERCP): diagnostic usefulness. Gastroenterol Hepatol
1998;21(4):174 80.
Sarles HG. Chronic calcifying pancreatitis-chronic
alcoholic pancreatitis. Gastroenterology 1974;66:
604 16.
Van Hoe L, Gryspeerdt S, Ectors N, et al. Nonalcoholic duct-destructive chronic pancreatitis: imaging
findings. Am J Roentgenol 1998;170:643 7.
Nealon WH, Townsend Jr CM, Thompson JC. Preoperative endoscopic retrograde cholangiopancreatography (ERCP) in patients with pancreatic pseudocyst
associated with resolving acute and chronic pancreatitis. Ann Surg 1989;209:532 8.
Cooperman AM. Surgery and chronic pancreatitis.
Surg Clin North Am 2001;81(2):431 55.
American Society for Gastrointestinal Endoscopy.
ASGE guidelines for clinical application. The role
of ERCP in diseases of the biliary tract and pancreas.
Gastrointest Endosc 1999;50(6):915 20.
Izbicki JR, Bloechle C, Knoefel WT, Rogiers X,
Kuechler T. Surgical treatment of chronic pancreatitis
and quality of life after operation. Surg Clin North
Am 1999;79(4):913 44.
Arslan A, Buanes T, Geitung JT. Pancreatic carcinoma: MR, MR angiography and dynamic helical CT in
the evaluation of vascular invasion. Eur J Rad
2001;38(2):151 9.
Johnson PT, Outwater EK. Pancreatic carcinoma versus chronic pancreatitis: dynamic MR imaging. Radiology 1999;212:213 8.
Karawasa E, Goldberg HI, Moss AA, Federle MP,
London SS. CT pancreatogram in carcinoma of the
pancreas and chronic pancreatitis. Radiology 1983;
148:489 93.
Elmas N, Yorulmaz I, Oran I, Oyar O, Ozutemiz O,
Ozer H. A new criterion in differentiation of pancreatitis and pancreatic carcinoma: artery-to-vein ratio using the superior mesenteric vessels. Abdom Imaging
1996;21:331 3.
113
[81] Ichikawa T, Sou H, Araki T, Arbab AS, Yoshikawa T,

Ishigame K, et al. Duct-penetrating sign at MRCP:
usefulness for differentiating inflammatory pancreatic
mass from pancreatic carcinomas. Radiology 2001;
221:107 16.
[82] Freeny PC, Bilbao MK, Katon RM. Blind evaluation
of endoscopic retrograde cholangiopancreatography
(ERCP) in the diagnosis of pancreatic carcinoma:
the double duct and other signs. Radiology 1976;
119:271 4.
[83] Macken E, Drijkoningen M, Van Aken E, Van Steenbergen W. Brush cytology of ductal strictures during
ERCP. Acta Gastroenterol Belg 2000;63(3):254 9.
[84] Kim MJ, Mitchell DG, Ito K, Outwater EK. Biliary
dilatation: differentiation of benign from malignant
causesvalue of adding conventional MR imaging
to MR cholangiopancreatography. Radiology 2000;
214:173 81.
[85] Sheridan MB, Ward J, Guthrie JA, Spencer JA, Craven CM, Wilson D, et al. Dynamic contrast-enhanced
MR imaging and dual-phase helical CT in the preoperative assessment of suspected pancreatic cancer: a
comparative study with receiver operating characteristic analysis. Am J Roentgenol 1999;173:483 90.
[86] Gasslander T, Arnelo U, Albiin N, Permert J. Cystic
tumors of the pancreas. Dig Dis 2001;19(1):57 62.
[87] Yeo CJ, Bastidas JA, Lynch-Nyhan A, Fishman E,
Zinner M, Cameron J. The natural history of pancreatic pseudocysts documented by computed tomography. Surg Gynecol Obstet 1990;170:411 7.
[88] Pitchumoni CS, Agarwal N. Pancreatic pseudocysts.
When and how should drainage be performed? Gastroenterol Clin North Am 1999;28(3):615 39.
[89] Boggi U, Di Candio G, Campatelli A, Pietrabissa A,
Mosca F. Nonoperative management of pancreatic
pseudocysts. Problems in differential diagnosis. Int J
Pancreatol 1999;25(2):123 33.
[90] Lumsden A, Bradley 3rd EL. Pseudocyst or cystic
neoplasm? Differential diagnosis and initial management of cystic pancreatic lesions. Hepatogastroenterology 1989;36(6):462 6.
[91] Lee YT. Cystadenocarcinoma versus pseudocyst of
the pancreas: a difficult differential diagnosis. Curr
Surg 1989;46(3):202 6.
[92] Zamboni G, Scarpa A, Bogina G, et al. Mucinous
cystic tumors of the pancreas: clinicopathological features, prognosis and relationship to other mucinous
cystic tumors. Am J Surg Pathol 1999;23:410 22.
[93] Nagar E, Ueki T, Chijiiwa K, et al. Intraductal papillary mucinous neoplasms of the pancreas associated
with so-called mucinous ductal ectasia. Histochemical and immunohistochemical analysis of 29
cases. Am J Surg Pathol 1995;19:576 89.
[94] Compton CC. Serous cystic tumors of the pancreas.
Semin Diagn Pathol 2000;17:43 55.
[95] Irie H, Honda H, Aibe H, et al. MR cholangiopancreatographic differentiation of benign and malignant
intraductal mucin-producing tumors of the pancreas.
114
[96] Yamaguchi K, Ogawa Y, Chijiiwa K, Tanaka M. Mucin-hypersecreting tumors of the pancreas. Am J Gastroenterol 1992;87:634 8.
[97] Procacci C, Megibow A, Carbognin G, et al. Intraductal papillary mucinous tumor of the pancreas:
a pictorial essay. Radiographics 1999;19:1447 63.
[98] Procacci C, Graziani R, Bicego E, et al. Intraductal
mucin-producing tumors of the pancreas: imaging
findings. Radiology 1996;198:249 57.
[99] Megibow AJ, Lombardo FP, Guarise A, et al. Cystic
pancreatic masses: cross-sectional imaging observations and serial follow-up. Abdom Imaging 2001;
26:640 7.
[100] Agostini S, Choux R, Payan MJ, Sastre B, Sahel J,
Clement JP. Mucinous pancreatic duct ectasia in the
body of the pancreas. Radiology 1989;170:815 6.
[101] Azar C, Van De Stadt J, Rickaert F, et al. Intraductal
papillary mucinous tumours of the pancreas: clinical
and therapeutic issues in 32 patients. Gut 1996;39:
457 64.
[102] Raijman I, Kortan P, Walden D, Kandel G, Marcon
NE, Haber GB. Mucinous ductal ectasia: cholangiopancreatographic and endoscopic findings. Endoscopy 1994;26:303 7.
[103] Obara T, Maguchi H, Saitoh Y, et al. Mucin-producing tumor of the pancreas: natural history and serial
pancreatogram changes. Am J Gastroenterol 1993;88:
564 9.
[104] Sugiyama M, Atomi Y, Hachiya J. Intraductal papillary tumors of the pancreas: evaluation with magnetic
resonance cholangiopancreatography. Am J Gastroenterol 1998;93:156 9.
[105] Koito K, Namieno T, Ichiimura T, et al. Mucin-producing pancreatic tumors: comparison of MR cholangiopancreatography with endoscopic retrograde
cholangiopancreatography. Radiology 1998;208:
231 7.
[106] Onaya H, Itai Y, Niitsu M, Chiba T, Michishita N,

Saida Y. Ductectatic mucinous cystic neoplasm of the
pancreas: evaluation with MR cholangiopancreatography. Am J Roentgenol 1998;171:171 7.
[107] Fukukura Y, Fujiyoshi F, Sasaki M, et al. HASTE MR
cholangiopancreatography in the evaluation of intraductal papillary-mucinous tumors of the pancreas.
J Comput Assist Tomogr 1999;23:301 5.
[108] Yamaguchi K, Ogawa Y, Chiijiiwa K, et al. Mucinhypersecreting tumors of the pancreas: assessing the
grade of malignancy preoperatively. Am J Surg 1996;
171:427 31.
[109] Terris B, Ponsot P, Paye F, et al. Intraductal papillary
mucinous tumors of the pancreas confined to secondary ducts show less aggressive pathologic features as
compared with those involving the main pancreatic
duct. Am J Surg Pathol 2000;24:1372 7.
[110] Kobayashi G, Fujita N, Lee S, et al. Correlation between ultrasonographic findings and pathologic diagnosis of the mucin producing tumor of the pancreas.
Nippon Shokakibyo Gakkai Zasshi 1990;87:235 42.
[111] Obara T, Maguchi H, Saitoh Y, et al. Mucin-producing tumor of the pancreas: surgery or follow-up? Nippon Shokakibyo Gakkai Zasshi 1994;91:66 74.
[112] Sugiyama M, Atomi Y. Intraductal papillary mucinous tumors of the pancreas: imaging studies and
treatment strategies. Ann Surg 1998;228:658 91.
[113] Procacci C, Biasiutti C, Carbognin G, et al. Characterization of cystic tumors of the pancreas: CT accuracy.
J Comput Assist Tomogr 1999;23:906 12.
[114] Horvath KD, Chabot JA. An aggressive resectional
approach to cystic neoplasms of the pancreas. Am J
Surg 1999;178:269 74.
[115] Kanazumi N, Nakao A, Kaneko T, et al. Surgical
treatment of intraductal papillary-mucinous tumors
of the pancreas. Hepatogastroenterology 2001;48(4):
967 71.
Radiol Clin N Am 41 (2003) 115 144
MR angiography of the abdominal aorta

and peripheral vessels
Vincent B. Ho, MDa,*, William R. Corse, DOb
a
Department of Radiology, Uniformed Services University of the Health Sciences, 4301 Jones Bridge Road, Bethesda,
MD 20814, USA
b
MR Imaging, Doylestown Hospital, 595 West State Street, Doylestown, PA 18901, USA
MR angiography (MRA), in particular gadolinium

(Gd)-enhanced three-dimensional MRA, is well
suited for the evaluation of patients with suspected
or known disease of the abdominal aorta and peripheral vessels [1 3]. In most of these patients, the
underlying vascular disease is degenerative and associated with atherosclerosis. Atherosclerosis and its
related sequelae are leading causes for morbidity and
mortality in the Western world. Renal insufficiency is
prevalent in this population because of underlying
diabetes mellitus or renal artery disease. The ability
of MRA to provide reliable arterial depiction without
the need for the use of nephrotoxic contrast agents
(eg, iodinated contrast media) is one of the more
compelling arguments for the use of MRA in this
patient population.
Time-of-flight and phase-contrast MRA

The potential of MR imaging to illustrate arterial
structures noninvasively using time-of-flight (TOF)
and phase-contrast (PC) MRA has been known for
almost two decades [4 6]. These techniques rely on
The opinions or assertions contained herein are the private

views of the authors and not to be construed as official or
reflecting the views of the Uniformed Services University of
the Health Sciences or the Department of Defense.
E-mail address: vho@usuhs.mil or vho@nih.gov
(V.B. Ho).
the properties of moving protons (flowing blood) and

can be performed as either a two- or a three-dimensional acquisition. TOF MRA relies on flow-related
enhancement or in-flow effect caused by the entry
of unsaturated protons into the imaging slice (twodimensional) or volume (three-dimensional). Arterialto-background image contrast is improved by the
application of repetitive radiofrequency pulses to
suppress the signal from stationary background tissue. Arterial in-flow (and arterial signal) is highest if
blood flow is brisk; the imaging slice (or volume in
the case of three-dimensional TOF) is thin; and
imaging is performed perpendicular to the direction
of flow. Vertically oriented vessels, such as the aorta
and peripheral arteries, are best imaged using axial
TOF scanning. Unfortunately, the acquisition of
images perpendicular to the length of the vessel is
inefficient and can result in long acquisition times.
Axial two-dimensional TOF MRA of the peripheral
vessels (from the aortic bifurcation to the ankle) can
often require 2 or more hours to accomplish. Long
imaging times increase the likelihood of motion
artifacts not only from physiologic motion (eg, respiration and peristalsis) but also from bulk patient
movement. This is further complicated by the fact
that viewing of the vessels is best performed in their
long axis. The use of thicker slices or partitions to
shorten acquisition time has the untoward effect of
not only increased saturation concerns but also
decreasing the spatial resolution of the vessels on
the subsequent longitudinal image reformation used
for image interpretation.
The other traditional MRA technique is PC, which
relies on the phase shifts that protons experience
PII: S 0 3 3 8 - 3 8 9 0 ( 0 2 ) 0 0 0 6 2 - 3
116
V.B. Ho, W.R. Corse / Radiol Clin N Am 41 (2003) 115144
when they move along a gradient field. In practice,

PC MRA generally requires longer scan times and is
technically more challenging to perform than TOF
MRA and not widely used. For targeted imaging of
smaller vascular regions, however, such as the renal
arteries, PC MRA has been found to be useful,
especially as an adjunct to TOF MRA [7 9]. For a
PC MRA, the operator must set an appropriate flow
direction and velocity encoding, which in cases with
complex flow or geometries may be difficult to
determine a priori.
Because TOF and PC imaging rely on flow, both
techniques are prone to artifacts related to disruptions
or variations in blood flow (eg, pulsatile flow and
turbulent flow) [10,11]. Turbulent flow is particularly
problematic because it results in intravoxel phase
dispersion and ultimately in vascular signal loss. This
is common about stenoses and regions with complex
geometries, which unfortunately are typically the
regions of clinical interest. Intravoxel dephasing is a
known limitation of TOF and PC MRA, which results
in the overestimation of the severity of a stenosis or
even in the erroneous appearance of a stenosis or
occlusion. These pitfalls and the other previously
mentioned issues related to TOF and PC imaging
have contributed to the general lack of enthusiasm for

their routine clinical use for imaging of the aorta and
peripheral vessels.
Gd-enhanced three-dimensional MRA

In the early 1990s, a new MRA technique
for aortography called Gd-enhanced three-dimensional MRA was introduced by Prince et al [12].
Unlike TOF and PC MRA, this method does not
rely on blood flow but rather on the T1 shortening
effects of circulating Gd-chelate contrast media. The
technique is quick, fairly easy to perform, and provides
high-resolution three-dimensional arterial image sets.
Artifacts related to spin saturation, slow flow, or
turbulent flow encountered with TOF are minimal
for Gd-enhanced three-dimensional MRA. As a result,
Gd-enhanced three-dimensional MRA can be tailored
for most efficient high spatial resolution imaging
(ie, parallel to the length of the vessel).
On Gd-enhanced three-dimensional MRA, arterial
signal relies on the concentration of Gd within
the lumen of the vessel during image acquisition.
The resultant images are essentially lumingrams
Fig. 1. A 77-year-old man with hypertension. On standard coronal maximum intensity projection (MIP) (A) from a Gd-enhanced
three-dimensional MRA, the proximal renal arteries are noted to be normal and patent. On oblique coronal MIP (B) and axial
subvolume MIP (C), however, the occluded left upper pole segmental renal artery (large arrow) and the moderate stenosis (small
arrow) of the left lower pole segmental renal artery are better visualized.
117
Fig. 2. A 49-year-old man with rectal cancer. Gd-enhanced three-dimensional MRA of the aortoiliac vessels was performed as a
preprocedural road map for intra-arterial chemotherapy planning. The arterial anatomy is well seen on volume-rendered projection of the three-dimensional data set. (A) Coronal maximum intensity projection (MIP). (B) Sagittal MIP.
118
Fig. 3. An 82-year-old woman suspected of having

mesenteric ischemia. Coronal volume-rendered projection
of the Gd-enhanced three-dimensional MRA demonstrates
normal orientation of the celiac artery and superior mesenteric
artery. Multiplanar reformation (not shown) noted both
vessels to have a normal caliber.
similar to that of conventional x-ray angiography.

Gd-enhanced three-dimensional MRA provides
data that can be formatted in angiographic projections
identical to those of x-ray angiography (views familiar to referring surgeons) without the risks related
to catheterization, nephrotoxicity, and radiation exposure inherent to x-ray angiography. Gd-enhanced
three-dimensional MRA has the additional benefit
of providing volumetric (three-dimensional) angiographic data sets, which can be used for improved
projectional viewing. Arteries are often tortuous
or obscured by overlapping structures (eg, large
aneurysm) on conventional x-ray angiography. Gdenhanced three-dimensional MRA can provide not
only the standard angiographic projections but also
the opportunity to view arteries in nonstandard obliquities using maximum intensity projection (MIP).
Overlying structures can be removed easily using
subvolume MIP (Fig. 1) or multiplanar reformation
(MPR). The three-dimensional data can also be
processed for advanced viewing using volume rendering, shaded surface display, and virtual intraarterial endoscopy [13,14]. Direct volume rendering
(Figs. 2 4) often can provide a different perspective
and may be helpful not only for image interpretation
but also for preoperative surgical discussions and
conferences. Standard MIP and MPR, however,
are all that are typically needed for routine interpretation [13].
Gadolinium-enhanced three-dimensional MRA
has consistently been shown to be accurate and
preferable to traditional noncontrast MRA techniques
for evaluation of not only the aorta but also the
lower extremity arteries [1 3]. In many institutions,
Gd-enhanced three-dimensional MRA is steadily
emerging as the preferred method for evaluation
of the abdominal aorta, and the peripheral run-off
vessels. The growing popularity of Gd-enhanced
three-dimensional MRA has been facilitated by MR
scanner and equipment manufacturers who have
designed new pulse sequences, interactive timing
algorithms, improved user-interfaces, and coil products specifically for the performance of Gd-enhanced
three-dimensional MRA. In addition, a large variety
of vendor and third-party products are now available
for soft-copy interpretation and viewing of the threedimensional data sets. In the ensuing sections, the
theory and technical considerations associated with
Gd-enhanced three-dimensional MRA are discussed
followed by a brief clinical discussion of interpretative issues for several common clinical indications
for MRA of the abdominal aorta, its branches, and
peripheral vessels.
Principle of Gd-enhanced three-dimensional MRA

Arterial depiction is optimized by proper timing of
data acquisition, especially the low spatial frequency
Fig. 4. A 69-year-old woman with a celiac artery aneurysm.

The volume-rendered projection of a Gd-enhanced threedimensional MRA clearly illustrates an aneurysm of the
celiac artery (arrow).
k-space data (center of k-space) for the period of peak

arterial enhancement (ie, peak concentration of Gd)
[15 17]. Late data acquisition can result in substantial venous contamination of the image sets and can
hinder image interpretation. Premature image acquisition can result in insufficient arterial Gd and poor
vascular depiction. The acquisition of central k-space
data during the period of preferential arterial enhancement (arterial phase) is generally preferable. Gdenhanced three-dimensional MRA was originally
described using a slow venous infusion 40 to 60 mL
Gd-chelate contrast media during a long (3 to
4 minutes) three-dimensional spoiled gradient echo
acquisition, one typically used for three-dimensional TOF MRA, leading to its early description as
Gd-enhanced three-dimensional TOF MRA. The
slow contrast infusion (0.3 mL/second) prolonged
the arterial phase of enhancement and delayed significant venous enhancement, extending the time
window for preferential arterial enhancement (Fig. 5).
Improvements in gradient strength and pulse
sequence design have yielded faster data acquisition
speeds (eg, 20 to 30 seconds per three-dimensional
acquisition, and recently with sensitivity encoding
119
[18] 10 to 20 seconds per three-dimensional acquisition). This has enabled the performance of breathhold
image acquisition, which minimizes respiratory
motion artifacts and significantly improves arterial
visualization of abdominal aortic branch vessels
[19 22]. Faster imaging, however, has necessitated
more accurate timing of data acquisition.
Timing
There are several methods for achieving proper
timing of a Gd-enhanced three-dimensional MRA
[15,19,20,23 26]. The simplest is using a fixed
timing delay (eg, 15 seconds). This can often be
unreliable, however, because circulatory times are
highly variable, especially if the patient has a poor
ejection fraction or large capacious aortic aneurysm.
The arrival of a contrast bolus in the abdominal aorta
can take from 10 to 60 seconds [19]. The preferred
methods are the use of a timing bolus injection
[19,20], a triggering algorithm [23,24,26], or a fast
multiphase technique [25]. The timing bolus strategy
entails the administration of a small 1- to 2-mL test
bolus at the same rate as the actual bolus (eg, 2 mL/
Fig. 5. Diagram of vascular signal intensity as it relates to bolus injection rate. Fast bolus injection results in higher arterial
contrast media concentrations and higher signal intensity. With faster injection rates, however, the duration of preferential arterial
enhancement is diminished because of earlier and more significant venous enhancement than seen with slower injection rates.
Slow injection rates prolong the arterial phase; however, the maximum arterial concentration of contrast media is lower. Very
slow injection rates may result in insufficient signal for adequate arterial visualization. (From Ho VB, Choyke PL, Foo TKF, et al.
Automated bolus chase peripheral MR angiography: initial practical experiences and future directions of this work-in-progress.
J Magn Reson Imaging 1999;10:376 88; with permission.)
120
second for a Gd-enhanced three-dimensional MRA of

the aorta) and imaging the target vessel at a high
enough temporal rate (eg, one to two frames per
second). For this technique it is critical that sufficient
flush (eg, 30 mL) is used to ensure that the bolus is
within the central vasculature and that the contrast
injections of the test bolus and actual bolus are
similar. Use of an MR imaging-compatible injector
(eg, Spectris, Medrad, Indianola, PA; and Optistar,
Mallinckrodt, St. Louis, MO) minimizes variations in
the delivery of contrast boluses. If monitoring is
performed in a plane perpendicular to the vessel
(eg, axial timing scan for the abdominal aorta),
superior and inferior saturation bands should be used
to minimize the in-flow signal and ensure vascular
signal is attributable only to the contrast bolus arrival.
The merit of this method is that it can be universally
performed on all current scanner platforms.
There also are several vendor-specific real-time

triggering options for Gd-enhanced three-dimensional
MRA. In one technique called automated bolus
detection algorithm (SmartPrep, GE Medical Systems, Waukesha, WI [23]), a monitoring volume is
placed in addition to the three-dimensional MRA
volume. The algorithm is able to detect the bolus
arrival into the monitoring volume and automatically
initiates the MRA data acquisition once the operatordefined thresholds are exceeded. There is a change in
scanner noise between the monitoring phase and the
MRA data acquisition that provides an audible cue
for the coordination of the patients breathholding.
Another real-time triggering algorithm uses MR
fluoroscopy (BolusTrak, Philips Medical Systems,
Best, The Netherlands; Care Bolus, Siemens Medical
Systems, Iselin, NJ). This technique provides a MR
fluoroscopic image of the target vasculature and
Fig. 6. Proper alignment of preferential arterial-phase enhancement for a variety of k-space schemes used for Gd-enhanced threedimensional MRA. The critical issue for all the schemes is for the central k-space data (ie, low spatial frequency data) to be
acquired during the plateau phase of arterial enhancement. In the conventional sequential k-space scheme, the central k-space
data are acquired during the middle of the data acquisition period. In both the conventional centric and elliptical centric
acquisition schemes, the central k-space data are obtained at the beginning of imaging. Note that with conventional centric
acquisitions, k-space is only centric in ky and that the high spatial frequency encodings in kz are also acquired during each linear
pass and the central k-space encodings in ky and kz are gathered more efficiently (ie, acquired more quickly) in the elliptical
centric acquisition scheme. Partial Fourier imaging with reverse sequential acquisition ordering can also provide a compact
acquisition of low spatial frequency data during the beginning of image acquisition. Note that low spatial frequency data are best
obtained during the plateau period of arterial enhancement. Acquisition of central k-space data prematurely during the rapid rise
in arterial signal (open arrow) can result in significant ringing artifacts (see Fig. 7). (Adapted from Ho VB, Foo TKF, Czum JM,
et al. Contrast-enhanced magnetic resonance angiography: technical considerations for optimized clinical implementation. Top
Magn Reson Imaging 2001;12:283 99; with permission.)
enables the operator to trigger the MRA data acquisition manually on contrast bolus arrival [26].
The final timing method is simply to perform
multiple fast MRA acquisitions in succession [25].
This tact, also known as multiphase or time-resolved
imaging, assumes that at least one of the MRA
acquisitions is performed properly during the arterial
phase of the bolus. The typical compromise for high
temporal resolution (5 to 8 seconds per three-dimensional acquisition) is lower spatial resolution of any
individual acquisition. This method may have little
use in patients with a poor breathholding capacity
because the limitations in breathholding may preclude the acquisition of sufficient data sets during a
single breathhold to ensure arterial-phase imaging.
Furthermore, respiratory motion during these acquisitions significantly degrades what are already lower
spatial resolution data sets.
Pulse sequence
Gadolinium-enhanced MRA is traditionally performed with a T1-weighted fast three-dimensional
spoiled gradient echo pulse sequence using the shortest possible repetition time (TR) and echo time (TE)
to ensure the fastest possible imaging speed. The use
of a three-dimensional acquisition provides high
121
spatial resolution and improves background suppression. Radiofrequency spoiling and the use of a higher
flip angle improve the T1 weighting of the acquisition
and the arterial signal following contrast administration. The imaging parameters should be tailored to
afford the highest possible spatial resolution for the
allotted time period, which for a breathhold acquisition is generally 20 to 30 seconds. Prescription of a
volume with a matrix of 256 224 to 256, partition
thickness of 1.5 to 2.5 mm, and 40 to 60 partitions is
usually sufficient for imaging the abdominal aorta
during a 20- to 30-second breathhold Gd-enhanced
three-dimensional MRA of the abdominal aorta.
Knowledge of the k-space trajectory of the threedimensional pulse sequence is also critical for proper
timing [15]. Historically, Gd-enhanced three-dimensional MRA had only been implemented using a
traditional sequential k-space scheme in which the
k-space is filled linearly in a sequential fashion from
top to bottom with the low spatial frequency data
(center of k-space) being acquired during the middle
of the imaging period. Because the central k-space
data are responsible for most image contrast, its
acquisition should be timed for peak arterial enhancement, and preferably before significant venous
enhancement occurs (Fig. 6). With the development
of real-time triggering methods, however, a variety of
Fig. 7. Ringing artifact on breathhold renal Gd-enhanced three-dimensional MRA in a 36-year-old man with left renal artery
stenosis. This artifact is recognized by the presence of bright and dark lines ([A] coronal maximum intensity projection, [B]
coronal source image) that parallel the edge of the enhancing abdominal aorta (small arrows) and results from the premature
acquisition of low spatial frequency data during leading edge of the contrast bolus when arterial signal is rapidly rising. This
artifact is more common with centric acquisition ordering, which acquires central k-space data early. Ringing artifact can be
avoided by timing for the low spatial frequency to be obtained during the plateau phase of the arterial enhancement (see Fig. 6).
Note that despite the artifacts, the patients left renal artery stenosis (large arrow) was well delineated. (Adapted from Ho VB,
Foo TKF, Czum JM, et al. Contrast-enhanced magnetic resonance angiography: technical considerations for optimized clinical
implementation. Top Magn Reson Imaging 2001;12:283 99; with permission.)
122
alternate k-space schemes were designed. In these

alternate k-space schemes, the central k-space data
are acquired during the beginning of the imaging
period, which improves the ability to synchronize the
bolus arrival with the critical central k-space data
acquisition. This allows monitoring of contrast arrival
within the target vessel itself versus proximal or
upstream. Two such alternate k-space schemes are
centric-phase ordering [23] and elliptical centricphase ordering (see Fig. 6) [26]. Partial Fourier
imaging (eg, 0.5 NEX) is another alternate method
for selective k-space sampling. As with traditional

sequential phase-ordered acquisitions, partial Fourier
imaging uses a linear k-space sampling scheme but
only a little over a half of k-space is acquired. As
such, the center of k-space can be prescribed to be
acquired either at the end (conventional sequential) or
at the beginning (reverse sequential) of the acquisition. The disadvantage of partial Fourier imaging is a
decrease in signal-to-noise, but with Gd-enhanced
three-dimensional MRA there is typically sufficient
arterial signal if timing is proper. With these alternate
Fig. 8. A 72-year-old man with a 9-cm abdominal aortic aneurysm. The extent of this large aortic aneurysm is displayed on the
arterial-phase Gd-enhanced three-dimensional MRA ([A] coronal maximum intensity projection [MIP], [B] sagittal MIP) and
delayed-phase Gd-enhanced three-dimensional MRA ([C] coronal MIP). Note the improvement in visualization of the infrarenal
aorta on the later delayed-phase acquisition (C) compared with the arterial-phase images (A,B). This is secondary to the slow
aortic flow within the large abdominal aortic aneurysm. Performing two acquisitions (arterial phase and delayed phase) after the
administration of contrast agent is prudent because it is hard to know a priori whether the patient has slow blood flow.
Furthermore, the delayed-phase images can often provide diagnostic quality angiographic images should there be inadequate
patient breathholding or motion during the arterial-phase acquisition. An axial image (D) from a late delayed-phase axial twodimensional spoiled gradient echo acquisition (ie, traditional axial two-dimensional time-of-flight MRA) taken through the
abdominal aorta delineates the circumferential mural thrombus (T) within the aneurysm and provides a better assessment of
actual aortic wall-to-wall diameter (arrows). (Adapted from Ho VB, Prince MR, Dong Q. Magnetic resonance imaging of the
aorta and branch vessels. Coron Artery Dis 1999;10:141 9; with permission.)
k-space sampling schemes, it is particularly important

not to acquire central k-space views during the rapid
rise in arterial Gd (see Fig. 6) because this can result
in a ringing artifact [16] seen as alternating black and
white lines about the edges of arteries (Fig. 7).
123
Bolus delivery
Faster pulse sequences have enabled the achievement of high-quality Gd-enhanced three-dimensional MRA more efficiently using smaller doses
Fig. 9. A 66-year-old man with a small infrarenal abdominal aortic aneurysm (AAA). On the coronal maximum intensity
projection (MIP) (A), a small AAA can be seen (arrow) well below the renal arteries, which are noted to be solitary for each
kidney and to have a normal caliber. On sagittal subvolume MIP (B), the ventral origins of the celiac artery (thick arrow) and the
superior mesenteric artery (thin arrow) are noted to also have a normal caliber. The contour and shape of the lumen of the small
infrarenal AAA (arrow) is particularly well seen on a volume-rendered projection (C).
124
of Gd-chelate contrast media [24,27,28]. With their

shorter bolus length requirements, faster imaging
provides the opportunity to deliver the contrast
media at a higher injection rate, which can provide
sufficiently high arterial Gd concentrations using
much lower doses (see Fig. 5). In general, contrast
injections are best delivered by a right antecubital
vein using a 22-guage or larger angiocatheter. The
reliability of contrast administration can be
improved by the use of an MR-compatible injector
and is particularly important for those who choose a
test bolus for timing. In general, contrast media
injections should always be accompanied by a
sufficiently large saline flush to ensure that the
contrast bolus is pushed out of the tubing set and

peripheral venous system and well into the central
circulation. Recently, Boos et al [29] demonstrated a
clear advantage for the use of a 30-mL or larger
saline flush for Gd-enhanced three-dimensional
MRA in that it significantly prolonged the duration
of arterial enhancement.
MRA of the abdominal aorta

An MRA of the abdominal aorta is primarily
performed for the assessment of aortic aneurysm;
aortic dissection; aortic occlusion; or a suspected
Fig. 10. A 75-year-old woman with hypertension and an infrarenal AAA. On coronal maximum intensity projection (MIP) (A) a
long, fusiform infrarenal AAA that extends to the aortic bifurcation. High-grade stenoses are also noted at the origins of both
common iliac arteries. On an oblique coronal subvolume MIP (B), a high-grade stenosis of the left renal artery is noted (arrow).
A high-grade stenosis (arrow) was also noted in the right renal artery at its origin; however, this was best seen on an oblique axial
subvolume MIP (C) from below.
branch vessel stenosis (especially renal artery stenosis). As discussed in the ensuing sections, breathhold Gd-enhanced three-dimensional MRA can
adequately answer the clinical questions related to
these aortic diseases. Gd-enhanced MRA of the
abdominal aorta is best performed in the coronal or
oblique coronal three-dimensional prescription. On
occasion, a sagittal acquisition may be preferable for
imaging arterial branches that originate ventrally,
such as the superior and inferior mesenteric arteries,
125
especially if the patients breathholding ability is

limited and the volume needs to be minimized
because of imaging time considerations. The superior
mesenteric artery, however, can often be well visualized on a sagittal reconstruction of a coronal threedimensional MRA data set if the vessel is included
within the field of view. The examination, however,
should always include an axial T1-weighted fast spin
echo and an axial T2-weighted fast spin echo, preferably performed before the MRA. These initial
Fig. 11. A 78-year-old man with chronic type B aortic dissection. On oblique sagittal maximum intensity projection (A), the
patient is noted to have a very tortuous thoracic aorta with only a hint of the dissection, which begins in the distal arch beyond the
subclavian artery (not shown). On oblique sagittal multiplanar reformation (MPR) (B), however, spiral extension of the intimal
tear into the abdominal aorta is clearly seen. Oblique axial MPRs at the levels of the celiac artery (C), superior mesenteric artery
(D), and renal arteries (E) demonstrate that all four arteries originate from the true lumen (T ) and not the false channel (F ).
126
sequences enable the evaluation of the aortic diameter

and wall and screening for any unexpected visceral
pathology (eg, pelvic kidney).
In general, before the breathhold Gd-enhanced
three-dimensional MRA, an unenhanced breathhold
three-dimensional MRA using the same imaging
parameters should be performed as a trial run. This
not only ensures that the three-dimensional volume is
appropriately placed and artifacts, such as phase
wrap, do not overlap critical regions but also provides
a preparatory experience for the patient (and the
operator) in which he or she can familiarize themselves with the breathholding requirements.
should always include at least two postcontrast
acquisitions (arterial phase and delayed phase). The
additional time is trivial (additional breathhold) and
the second acquisition (delayed-phase MRA) often
can provide additional information. If blood flow is
slow as in a large aneurysm (Fig. 8), the initial arterialphase MRA may be completed before the attainment
of sufficient Gd concentrations have been achieved

within the distal aorta. Similar concerns arise in an
aortic dissection where flow within the false channel
is often slow. Imaging multiple phases has additional
benefits for renal artery evaluations because it also
enables the detection of delays or changes in parenchymal enhancement [30]. An additional benefit for
delayed-phase imaging is not only to provide a
secondary set of images but an improved depiction
or mural thrombus. The routine performance of an
axial two-dimensional fast spoiled gradient echo pulse
sequence (ie, a conventional axial two-dimensional
TOF MRA) after the coronal breathhold acquisitions
can be a helpful option to consider for the evaluation
of thrombus (see Fig. 8).
A contrast dose of 20 to 30 mL of Gd-chelate
injected at 2 mL/second [21] usually provides sufficient illustration of the abdominal aorta on Gdenhanced three-dimensional MRA. Although a single
dose (20 mL [27]) of a Gd-chelate contrast agent has
been shown to be adequate for renal artery imaging, a
Fig. 12. A 60-year-old woman with chronic type B aortic dissection but worsening vague abdominal pain. Sagittal maximum
intensity projection (MIP) (A) from a sagittal Gd-enhanced three-dimensional MRA demonstrates a narrowing of the celiac
artery (arrow) at its origin. On an oblique axial subvolume MIP (B), the extension of the dissection into the proximal celiac
artery is better visualized. Note that the true lumen is bright but the false channel was thrombosed and only apparent by its mass
effect on the true lumen (arrows) on Gd-enhanced three-dimensional MRA.
larger dose of 30 mL (or 0.2 mmol/kg) is recommended in patients with a large abdominal aortic
aneurysm (AAA), an aortic dissection, or aortic
occlusion because this ensures a sufficiently high
arterial Gd concentration for adequate visualization
of the arterial structures.
Clinical considerations
Abdominal aortic aneurysm
Aneurysms are defined as enlargement of the
arterial diameter by 50% or more from its normal
caliber, which for the abdominal aorta is generally
greater than or equal to 3 cm [31]. AAAs are common
especially in men above the age of 55 and in women
above the age of 70 [32]. The urgency of this
diagnosis relates to its risk of rupture, which is fatal
in most cases (81% to 94% [33]). AAAs are frequently asymptomatic, however, until they rupture.
Large AAAs (greater than 5 cm) have a 25% to
41% likelihood of rupture within 5 years [34,35] and
generally are repaired surgically. The risk of rupture
of small AAAs (aortic diameter less than 4 cm), on
the other hand, is low (0% to 2% [34,35]). Because
the 30-day operative mortality risk for elective AAA
repair is 5% to 6% [36], AAAs with diameters less
than 4 cm are typically followed with periodic
surveillance to check for interval expansion, which
is typically 0.2 to 0.4 cm per year [34,35]. Of course,
rapid expansion of an AAA (eg, greater than 1 cm per
year), widening of the pulse pressure, or the mani-
127
festation of symptoms (eg, abdominal tenderness or

pain) favors early surgical intervention. Ideally, surgical repair is elective because emergent repair carries
roughly a 10-fold increase in operative morbidity
(30-day operative mortality of 40% to 50% [36]).
The elective repair of AAA with diameters between
4 and 5 cm, however, continues to be debated because rupture rates are 3% to 12% after 5 years [34].
Improvements in aortic stent grafts has enabled
successful endovascular repair of some aneurysms,
especially small infrarenal AAA. Although preprocedural assessments can be obtained using MRA,
magnetic susceptibility artifacts can be significant
for certain stent materials and CT generally has been
the modality of choice for imaging prestenting and
poststenting patients with AAA. There are a growing
number of stent grafts, however, which are made of
materials, such as nitinol or polytetrafluoroethylene,
that seem to have minimal artifacts on Gd-enhanced
three-dimensional MRA [37,38].
Degenerative aneurysms of the aorta are commonly associated with atherosclerosis but recent
observations suggest a multifactorial causation and
a categorization of most AAAs as nonspecific
[31,34]. AAA is typically fusiform but may on
occasion be saccular. A saccular aneurysm should
lead one also to entertain an infectious etiology (ie,
mycotic aneurysm) [39]. Aneurysms are typically
infrarenal (84%) but can also involve the entire
abdominal aorta (4%) or be limited to the pararenal
region (12%) [40]. Aneurysms of the proximal
Fig. 13. A 56-year-old man with Leriches syndrome who presented with hypertension and buttock claudication. Preoperative
Gd-enhanced three-dimensional MRA ([A] coronal maximum intensity projection [MIP]) demonstrates the characteristic
occlusion of the distal abdominal aorta below the renal arteries. A high-grade stenosis was also noted in the proximal left renal
artery (arrow). The postoperative Gd-enhanced three-dimensional MRA ([B] coronal MIP) demonstrates the aortobifemoral graft
that included revascularization of the left renal artery at the proximal anastomosis. (Courtesy of Qian Dong, MD, and Martin
Prince, MD, PhD, Ann Arbor, MI.)
128
abdominal aorta carry a higher operative morbidity

and are particularly challenging because aortic reconstruction requires proper incorporation of branch
vessel ostia (ie, renal arteries, celiac arteries, or
superior mesenteric artery).
Conventional T1-weighted spin echo images have
been found to be excellent for the delineation of
aortic dimensions [41,42]. Preoperative planning,
however, also requires not only the assessment of
aneurysm size (width, depth, and length) but also its
proximal and distal extent and its relationship to the
visceral branch vessels. In addition, evidence for
rupture, complications of the aortic wall, supernumerary renal arteries, obstructive disease of renal, celiac,
or mesenteric vessels, or an anomaly, such as a
horseshoe kidney, can significantly alter the surgical
plan [43]. Gd-enhanced three-dimensional MRA
(Figs. 9, 10) can illustrate these features accurately
and reliably and has been shown to be sufficient for
preoperative planning for AAA interventions
[22,44 47]. For example, Gd-enhanced three-dimensional MRA has been shown to predict correctly the
proximal anastomotic site for AAA repair in 95% of
patients, which was comparable with that of conven-
Fig. 14. A 50-year-old woman with hypertension. On Gd-enhanced three-dimensional MRA ([A] oblique coronal subvolume
maximum intensity projection (MIP), [B] axial subvolume MIP, [C] coronal volume-rendered projection, [D] coronal transparent
volume-rendered projection), the string of beads appearance (arrows) characteristic for fibromuscular dysplasia is noted in the
right renal artery, which looks comparable with that of conventional x-ray angiography (E). Note that the beaded appearance was
well seen in the right renal artery on Gd-enhanced three-dimensional MRA, especially on the volume-rendered projections (C,D).
(F) Mild fibromuscular dysplasia was also noted on conventional x-ray angiography in the proximal left renal artery (arrow).
This was suggested on Gd-enhanced three-dimensional MRA (G) but less clearly seen, most probably secondary to the inherent
lower spatial resolution of MRA.
tional x-ray angiography (97%) [46]. MIP and MPR

are particularly helpful for the identification of stenotic branch vessels, which are preferably revascularized at the time of the AAA repair. Gd-enhanced
three-dimensional MRA has been shown to be accurate for the detection of significant occlusive celiac,
renal, mesenteric, or iliac arterial disease (94% sensitivity and 98% specificity [47]).
Aortic dissection
Dissections much more frequently arise in the
thoracic aorta but can often extend inferiorly into the
abdominal aorta to involve the renal, celiac, mesenteric, and iliac arteries [48]. The principle concern
with dissections is the involvement of visceral
branch vessels, which can result in their obstruction.
On Gd-enhanced three-dimensional MRA, the true
and false channels and entry and exit intimal tears
can be well illustrated (Figs. 11, 12) [49 51]. MPR
129
of the three-dimensional data sets enables the selective viewing of individual aortic branch vessels and
the identification of their blood supply (ie, from true
versus false channel). The extension of an intimal
tear into the abdominal aorta typically spirals posterior laterally about the arch with the false channel
coursing to the left of the aorta potentially to involve
the left renal artery and possibly the celiac and
superior mesenteric arteries. Delayed-phase imaging
is recommended (ie, at least two postcontrast MRA
acquisitions) because flow within the false channel
may be slow and not adequately fill with contrast
media during the initial acquisition.
Aortic occlusion (Leriches syndrome)
Occlusion of the abdominal aorta is uncommon
but worth mentioning because MRA can be very
useful in this condition [43,52]. Abdominal aortic
occlusion may occur as a result of a variety of
Fig. 14 (continued )
130
causes, in the acute setting most commonly as a

result of embolism. Chronic occlusion most commonly results from thrombosis superimposed on
severe atherosclerotic involvement of the distal
abdominal aorta and common iliac arteries. Chronic
occlusion can produce Leriches syndrome (named
after Leriche who was the first to describe aortic
occlusion in 1940 [52]). Leriches syndrome refers to
the clinical syndrome that results from occlusion of
the infrarenal aorta. Patients typically have buttock
and thigh claudication, erectile impotence, atrophy of
the thigh musculature, and diminished femoral
pulses. Invariably patients with chronic occlusion
develop a rich collateral circulation. Arterial access
is limited in these individuals and Gd-enhanced
three-dimensional MRA (Fig. 13) can often be sufficient for the primary assessment of the occlusion or
the assessment of graft repairs.
Branch vessels
Renal artery stenosis. Although all major abdominal aortic branches can be well evaluated with
MRA, the renal arteries merit special discussion
because renal artery stenosis is particularly common
among patients with conditions affecting the aorta
and peripheral vessels. Up to 22% of patients with
infrarenal AAA (see Fig. 10) [53] and 45% of patient
with peripheral vascular disease [54] also have renal
artery stenosis. Renal artery stenosis frequently manifests clinically as systemic hypertension (70% [55]),
which can often be reversed with renal revisualization by balloon angioplasty, stenting, or vascular
surgery. Patients with renal artery stenosis can often
progress to end-stage renal disease if left untreated.
Atherosclerotic renal artery stenosis has a 35%
cumulative incidence of disease progression at
3 years and 51% at 5 years [56]. Renal artery
stenosis and disease progression are particularly
prevalent in diabetic patients, which comprise
roughly 50% of patients with renal artery stenosis
[55,56].
The preoperative identification of renal artery
stenosis is important and may augment or change
the surgical plan. Concomitant renal revascularization
during surgery for an AAA or aortoiliac occlusive
disease has been shown to result in significant
improvement or reversal of hypertension in most
patients [57,58].
The critical technical issue for achieving diagnostic renal MRA is spatial resolution, which ideally
is less than 1.5 mm in any single dimension [59].
The diagnosis that is especially challenging by Gdenhanced three-dimensional MRA is that of fibro-
muscular dysplasia because the changes may be

subtle (Fig. 14). Like atherosclerotic renal artery
stenosis, fibromuscular dysplasia can result in reversible systemic hypertension. Patients with fibromuscular dysplasia, however, are typically young
women; whereas, atherosclerosis tends to occur in
older men [60]. Fibromuscular dysplasia typically
has a string of beads appearance, which may be
subtle on Gd-enhanced three-dimensional MRA
(see Fig. 14).
has been shown to be very accurate (sensitivity 91%
to 100%, specificity 89% to 100% [30,61 66])
for the detection of greater than 50% diameter
stenoses of the main renal artery. The supplementation of Gd-enhanced three-dimensional MRA with a
postcontrast PC three-dimensional MRA can provide ancillary and often complementary information, which improves the specificity of MRA for
the detection of renal artery stenosis [64,67,68]. On
PC MRA, spin dephasing invariably is present in
hemodynamically significant renal artery stenosis.
PC MRA relies on blood flow and the phase shifts
that it experiences while moving across a gradient
field. Because of the significant time requirements,
this technique was never popular for routine clinical applications. Like the previously discussed
flow-based technique of TOF, PC MRA is also
prone to flow-related artifacts, such as intravoxel
dephasing about regions of arterial narrowing. After
contrast administration, however, arterial signal on
PC MRA is especially high [69]. Because the
technique is still sensitive to turbulent flow, intravoxel dephasing is still present on postcontrast
imaging and can be used to confirm the presence
of a hemodynamically significant stenosis (Fig. 15).
The performance of a Gd-enhanced three-dimensional MRA first provides a road map for the
appropriate prescription of the phase-contrast
MRA over a more limited anatomic region and a
more time-efficient PC acquisition.
Renal transplant evaluation. Without the concerns
of nephrotoxicity associated with CT and conventional x-ray angiography, Gd-enhanced three-dimensional MRA can be a good method for the
postoperative assessment of renal transplant recipients (Fig. 16) [70,71]. Dual-phase Gd-enhanced
three-dimensional MRA easily can assess the
patency of the vascular anastomoses of the transplanted kidney.
can also be used to screen potential renal donors
and has been found to be comparable with CT
131
Fig. 15. A 71-year-old man with a history of hypertension and diabetes mellitus. Gd-enhanced three-dimensional MRA ([A]
coronal maximum intensity projection) shows severe renal artery stenosis bilaterally (arrows). On phase-contrast threedimensional MRA (B), signal loss distal to the renal artery stenoses (arrows) is seen. This suggests that both arterial narrowings
are hemodynamically significant. Bilateral high-grade stenoses (75% on right and 80% on left) are noted on conventional x-ray
angiography (C). (Adapted from Hood MN, Ho VB, Corse WR. Three-dimensional phase contrast MR angiography: a useful
clinical adjunct to gadolinium-enhanced three-dimensional renal MRA? Mil Med 2002;167:343 9; with permission.)
angiography [72,73]. The critical issue for the preoperative evaluation of potential donors is to determine the most suitable kidney for expedient and safe
removal [74,75]. Imaging is performed to identify the
number of renal arteries, the presence of early branching arteries, unsuspected renovascular disease, or any
parenchymal disease (eg, renal cell carcinoma) that
may influence the choice of kidney. On dual-phase

Gd-enhanced three-dimensional MRA, renovascular
anatomy and anomalies (eg, renal ectopia and retroaortic or circumaortic renal vein) readily can be
identified. Supernumerary renal arteries (Fig. 17)
are particularly common (27% of kidneys [75]) and
although not a contraindication for renal donation
132
Fig. 16. A 48-year-old man with autosomal-dominant polycystic kidney disease. On Gd-enhanced three-dimensional MRA ([A]
coronal maximum intensity projection [MIP], [B] oblique coronal subvolume MIP, [C] coronal multiplanar reformation [MPR])
a normal and patent arterial anastomosis (arrow) of the transplant kidney (t) with the external right iliac artery (a) is noted. On
MPR (C), overlapping signal from the right external iliac artery could be removed, enabling improved visualization of the
anastomosis (arrow).
may affect the choice of kidneys for transplant or the

surgical approach.
MRA of the peripheral vessels

The primary indication of peripheral angiography
is for the evaluation of patients with suspected or
known peripheral arterial occlusive disease (PVOD).
Once again, patients typically have atherosclerosis.
Atherosclerotic PVOD is common and its prevalence

increases with age, affecting 20% of the population
over the age of 75 years, and is twice as common in
men [76,77]. Patients typically present with intermittent claudication in calf, thigh, or buttocks, which is
exacerbated with exercise or ambulation. Claudication is indicative of a diminished arterial flow reserve
and an inability to augment blood flow for the
increased metabolic demands of exercise. Symptoms
are typically self-limiting but can significantly impact
an individuals quality of life. It is this latter effect on
133
Fig. 17. Arterial-phase renal Gd-enhanced three-dimensional MRA using an automated bolus detection scheme that was
prescribed to monitor signal in a 3 3 3 cm volume within the mid-abdominal aorta at the level of the renal artery origins.
The breathheld renal Gd-enhanced three-dimensional MRA ([A] coronal maximum intensity projection [MIP], [B] oblique
subvolume MIP) in this 46-year-old male renal donor demonstrates supernumerary renal arteries (two right and three left renal
arteries). (Adapted from Ho VB, Foo TKF, Czum JM, et al. Contrast-enhanced magnetic resonance angiography: technical
considerations for optimized clinical implementation. Top Magn Reson Imaging 2001;12:283 99; with permission.)
quality of life that may bear significantly on the

decision to treat patients with intermittent claudication more aggressively [78 80].
In more severe cases of PVOD, ischemia may be
limb threatening and therapeutic intervention (eg,
balloon angioplasty, stenting, bypass graft placement,
or amputation) is typically required. These patients
typically complain of claudication at rest or develop
nonhealing ulcers or even gangrene in the affected
leg. The therapeutic option depends on the location of
the disease, degree of stenosis, and length of the
lesion. Focal stenosis or occlusion (length less than
5 cm) of iliac artery, for example, responds well to
balloon angioplasty or stenting [81 84]. Long iliac
artery stenoses or occlusions, however, have lower
long-term patency success rates and often require a
surgical bypass procedure. The decision to perform a
bypass procedure is always considered carefully
because failure of the bypass graft may result in a
higher level of amputation than initially required
[85,86].
Multistation Gd-enhanced three-dimensional MRA
(bolus-chase MRA)
Arteriography of patients with PVOD has been
particularly challenging because the length of the
vascular anatomy that must be illustrated (from at
least the aortic bifurcation to the level of the ankle or
distal trifurcation vessels) is extensive (eg, greater

than 1 m). This is required because lesions are
typically multiple and tandem lesions are common
(70%) [79]. Surgical planning requires comprehensive evaluation of the entire arterial territory. Repair
of a popliteal artery stenosis, for example, may have
little effect if the patient has an ipsilateral iliac
occlusion and generally in-flow disease (aortoiliac)
is treated first. Arteriography is also necessary to
illustrate potential donor and recipient sites for potential bypass.
Using individual fields of view of roughly 40 to
50 cm, peripheral MRA typically requires the
imaging of three or more overlapping locations or
stations. Using two-dimensional TOF technique, peripheral MRA has been successful at detecting arterial
stenoses greater than 50% (eg, sensitivity 85% to
92%, specificity 81% to 88% [87,88]). Aside from its
obvious clinical benefits versus x-ray angiography,
two-dimensional TOF MRA has also been shown
to demonstrate suitable infrapopliteal bypass recipients not visible on conventional x-ray angiography
(so-called occult run-off vessels), which are critical
for bypass graft planning [87,89,90]. The lengthy
time requirements (often greater than 2 hours) and
numerous pitfalls, however, have limited the acceptance of two-dimensional TOF imaging for routine
clinical peripheral MRA. Gd-enhanced three-dimensional MRA is much faster than two-dimensional
134
TOF MRA and has been found to afford improved

diagnostic performance for imaging the peripheral
vessels [1,2].
The most recent development of Gd-enhanced
three-dimensional MRA has been the development
of a technique called bolus-chase MRA [91 102].
Bolus chasing has been used for many years in
conventional x-ray angiography. The basic concept is to synchronize imaging with the arterial transit
of a single contrast bolus. In MR imaging, this
can be achieved by aligning table translation
with the arterial phase of an intravenously administered Gd-chelate contrast bolus (Fig. 18). Typically, a
40-mL or 0.2-mmol/kg dose of Gd-chelate contrast
media is administered at a slow rate (0.3 to 0.8 mL/
second) [93,94,97]. The rate of contrast infusion
should be adjusted so that the length of the contrast
bolus duration matches roughly the time required to
acquire the critical k-space data for the three overlapping stations (Fig. 19). For example, if imaging
requires 100 seconds (30 seconds per station with
5 seconds between stations), a 40-mL dose injected at
0.4 mL/second results in a 100-second bolus duration. Recently, a biphasic injection rate (Fig. 20) has
also been shown to be effective [101].
To ensure that the bolus duration and injection
rates are standardized, another tact is to dilute a
0.2-mmol/kg dose to a fixed volume (eg, 45 mL)
and inject it at a fixed rate (eg, 1 mL/second) [94].
The use of a 0.2-mmol/kg dose ensures the ability
to perform another Gd-enhanced three-dimensional
MRA using 0.1 mmol/kg should a segment require
additional investigation. The actual technique varies
with the imaging capabilities of the scanner that is
being used [95]. Some scanners, for example, are

capable of partial Fourier imaging (ie, 0.5 excitation or NEX) which allows for the foreshortening
of the bolus duration requirements (see Fig. 19).
Table translation can be performed manually by
simply unhooking the scanner table and sliding the
patient out of the bore [91 93,97]; by using a
specially designed coil on platform apparatus (eg,
SKIP, Magnetic Momments, Bloomfield, MI [101];
or AngioSURF, MR Innovation, Essen, Germany
[99]); or using software that integrates imaging with
automated table motion (eg, MoBI-Track, Philips
Medical Systems, Best, The Netherlands [96]; and
SmartStep, General Electric Medical Systems, Waukesha, WI [100]). The selection of table translation
method primarily depends on the individual MR
scanner because options vary between vendors. Each
vendor has similar variation in timing method options
(MR fluoroscopic trigger, automated bolus detection,
and so forth) and pulse sequence choice (partial
Fourier versus full Fourier, sequential versus centric
phase ordering, and so forth). Operators are advised
to familiarize themselves with the specific options
available with their scanner.
Although specific table motion technique and
timing methods may differ, the basic bolus-chase
MRA procedure remains fairly similar. All techniques
require a multistation localizer and matching precontrast coronal (or oblique coronal) three-dimensional
acquisitions at each location, typically using the same
table motion procedure as for the subsequent boluschase MRA. As with single-station Gd-enhanced
three-dimensional MRA, the precontrast multistation
three-dimensional MRA serves to ensure appropriate
Fig. 18. Schematic of multistation peripheral bolus chase three-dimensional MRA. Imaging of the peripheral vasculature requires
the imaging of three contiguous anatomic regions: the aortoiliac segment (station 1); the femoropopliteal segment (station 2); and
the tibioperoneal or trifurcation segment (station 3). (From Ho VB, Choyke PL, Foo TKF, et al. Automated bolus chase
peripheral MR angiography: initial practical experiences and future directions of this work-in-progress. J Magn Reson Imaging
1999;10:376 88; with permission.)
135
Fig. 19. Schematic of arterial-phase imaging of the contrast bolus at stations 1 through 3. The relative timing of the data
acquisition of the half-Fourier three-dimensional gradient echo sequences is diagrammed as A (station 1), B (station 2), and C
(station 3) with the center lines of k-space for each marked by diagonal lines. Note that the use of sequential view ordering for
station 1 and reverse sequential view ordering for station 3 results in a shortened duration required for central k-space coverage
during the arterial phase (ie, shortened critical arterial imaging period). (From Ho VB, Choyke PL, Foo TKF, et al. Automated
bolus chase peripheral MR angiography: initial practical experiences and future directions of this work-in-progress. J Magn
Reson Imaging 1999;10:376 88; with permission.)
anatomic coverage and familiarize the patient with

the procedure. For bolus-chase MRA, however, the
precontrast images have an additional benefit in that
they provide a mask for image subtraction, which can
significantly improve arterial visualization in the
peripheral vasculature, especially in the calf
[94,103,104]. It is advisable to take efforts to minimize bulk patient movement between the precontrast
and postcontrast imaging by minimizing not only the
time between acquisitions but also securing the
patients lower extremities whenever possible.
Bolus-chase MRA has been shown to depict
reliably and accurately peripheral arterial stenoses
greater than 50% (eg, sensitivity 81% to 95%; specificity 91% to 98% [93,97,98]). This technique can be
helpful not only for the initial screening of patients
with suspected PVOD (Figs. 20 22) but also for the
postoperative surveillance of graft patency (Fig. 23)
[91]. A known pitfall for the use of Gd-enhanced
three-dimensional MRA for postoperative evaluations
is its diminished ability to access stent graft patency
in patients who have undergone endovascular repair

with a stent that contains stainless steel (eg, Palmaz
stent) or cobalt-based alloy (eg, Wallstent). The
magnetic susceptibility from these stents results
in significant signal loss and precludes proper visualization of the stent lumen even on Gd-enhanced
three-dimensional MRA [37]. Stents that are made
of nitinol wire (eg, Cragg stent, Cragg Endo ProSystem 1, and Passager stent) and polytetrafluoroethylene (eg, Hemobahn stent), however, have been
shown to have minimal artifacts on Gd-enhanced
three-dimensional MRA [37,38]. In patients who
have undergone endovascular therapy using nitinolor polytetrafluoroethylene-based stent grafts, Gd-enhanced three-dimensional MRA can be a suitable
modality for the assessment of graft patency.
Current bolus-chase MRA methods are reliable
for imaging the peripheral vessels through the level
of the trifurcation. Visualization of the distal run-off
vessels, however, is often variable [15,94,102]. This
results from a combination of issues. Timing with
136
current techniques is only provided for the initial

station and timing for arterial-phase imaging of the
terminal station (third or fourth station) is understandably variable. In addition, imaging of the terminal station begins only after completion of imaging
for the proximal stations. These concerns are amplified in patients with PVOD because they often
have slow flow or significant intervening aneurismal
or occlusive disease in which the transit time for
contrast media can be highly variable and often

asymmetric. In patients with limb-threatening ischemia, the identification of recipient run-off vessels is
critical for successful bypass grafting. Visualization
of distal run-off vessels can be ensured by the
preliminary performance of a traditional two-dimensional TOF below the knee (especially foot). Most
patients with PVOD present with milder disease and
intermittent claudication, however, and therapy is
typically conservative (eg, smoking cessation and
regular exercise) and noninvasive [77]. In these
individuals, a three-station bolus-chase MRA typically can provide sufficient diagnostic information
for patient management.
Future directions
In addition to the use of higher field strength MR
scanners (eg, 3 T) and high performance gradients,
there are a variety of new techniques that may
significantly improve the speed of MRA data acquisition. New parallel imaging techniques, such as simultaneous acquisition of spatial harmonics [105] and
sensitivity encoding [18], use the spatial-encoding
properties of multiple phased-array coil elements to
reduce the number of requisite spatial-encoding
views. These can result in a significant reduction
(twofold or threefold) in scan time but at the cost of
signal-to-noise (approximately equal to the square
root of the scan time reduction factor). This is
especially promising for MRA of the abdominal aorta
Fig. 20. A 60-year-old man with atherosclerosis. Coronal

maximum intensity projection from a three-station Gd-enhanced bolus chase three-dimensional MRA using a biphasic
injection provided sufficient Gd for good visualization of the
abdominal aorta and iliac arteries (station 1); the femoropopliteal arteries (station 2); and the trifurcation vessels (station
3). The contrast was injected intravenously at 0.6 mL/second
for initial 20 mL and at 0.4 mL/second for the remaining
20 mL. This was followed by a saline flush at 0.4 mL/second. This provided a sufficiently long bolus to match the
100 seconds required for data acquisition (30 seconds for
each of three stations plus 5 seconds for each of the two table
movements between stations). Although, a single slow injection rate of 0.5 mL/second provides near equivalent bolus
duration, the slightly faster initial rate of injection provides a
higher concentration of Gd for improved visualization of the
abdominal aorta. The decrease in the injection rate for the
later half of the bolus ensures sufficient arterial Gd concentrations during the imaging of the distal trifurcation vessels. On this examination, fusiform dilatation of the distal
abdominal aorta and common iliac arteries and a moderate
stenosis in the distal right external iliac artery were noted.
Fig. 21. A 65-year-old man with right lower extremity

claudication. Coronal maximum intensity projection from a
three-station Gd-enhanced bolus chase three-dimensional
MRA demonstrates bilateral narrowing of the common iliac
arteries, which is much worse and high-grade in the right
common iliac artery (arrow). The remaining arterial segments were patent.
137
Fig. 22. A 53-year-old man with bilateral occlusion of the

proximal superficial femoral arteries. Coronal maximum
intensity projection from a three-station, Gd-enhanced bolus
chase three-dimensional MRA demonstrates not only the
occlusions but also the reconstitution of superficial femoral
artery at the mid-thigh level (arrows). Note the numerous
collateral vessels in the thigh about the regions of occlusion.
138
Fig. 23. Adult patient with atherosclerosis and a history of

abdominal aortic aneurysm (AAA) repair. The Gd-enhanced
bolus chase three-dimensional MRA demonstrates a residual
AAA in the proximal abdominal aorta but the aortobifemoral graft in the distal abdominal aorta is intact. Mild
irregularity is noted in the distal superficial femoral arteries
bilaterally consistent with mild to moderate disease.
where imaging speed can be used to acquire a larger

number of partitions for improved anatomic coverage
or more importantly higher spatial resolution. Alternatively, the speed can be used to achieve improved
temporal resolution and an increase in the number of
phases during a breathhold.
There also has been the introduction of hybrid
bolus-chase schemes. Maki et al [106] have demonstrated the feasibility of an interleaved image acquisition in which a two-dimensional MRA is performed at
the thigh station (station 2) between two three-dimensional MRAs (abdomen-pelvis, station 1; and calf,
station 3). This expedites imaging of the calf and can
minimize the concerns related to venous contamination. Another approach was described by Foo et al
[107] in which the three-dimensional acquisitions
were segmented such that only the central portions
of k-space (low spatial frequency data) are acquired
during an initial pass during the arterial phase of the
contrast bolus and remaining k-space data are
acquired later during a second pass during the delayed
phase. This technique called segmented volume
acquisition (shoot and scoot) enables more efficient
data acquisition because the time-critical portions of
k-space (ie, the center of k-space) are preferentially
acquired during the arterial phase of the bolus and
provides high spatial resolution data sets (Fig. 24).
Another innovation is time-resolved two-dimensional [102] and three-dimensional (eg, TRICKS
[108]) digital subtraction angiography. These techniques are somewhat similar to the multiphase threedimensional MRA scheme previously discussed
under timing methods but require specialized off-line
computer equipment or software, which are not yet
commercially available. Like multiphase threedimensional MRA, MR digital subtraction angiography may have limited use for imaging regions where
breathhold acquisitions are desired (ie, aortoiliac
region) but for imaging of the thigh, calf, and feet,
these techniques may be very helpful.
There have also been technical improvements
that relate to the pulse sequence. Until recently,
Gd-enhanced MRA has typically been performed
using a T1-weighted spoiled fast gradient echo pulse
sequence. Foo et al [109] recently reported success
using a steady-state free precession (TrueFISP, Siemens Medical System; FIESTA, General Electric
Medical Systems; balanced FFE, Philips Medical
Systems) for Gd-enhanced MRA. On steady-state
free precession, vascular signal is a function of the
ratio of tissue T2 to T1 relaxation times. This effect
can provide additional vascular signal contributions
that may improve luminal visualization despite low
Gd concentrations (Fig. 25).
The final developments have been in the contrast

agents [110]. There are a variety of new contrast agents
that have unique binding to large molecules like
albumin (eg, MS-325, Epix Medical, Cambridge,
MA) or inherently large structure (eg, NC100150,
Amersham Health, Buckinghamshire, United Kingdom) and are retained within the blood pool for
a prolonged period of time, whereas conventional
extracellular Gd-chelate contrast agents leak out
of the vessels within 2 to 3 minutes of venous
139
injection. Blood pool agents like conventional extracellular Gd-chelates rely on their T1-shortening effects
for improved signal on contrast-enhanced MRA. The
prolonged window of arterial signal improvement
affords a large temporal window for high spatial
resolution scanning. The main limitation of this technique is the significant venous enhancement that is
typically present after the initial 1 to 2 minutes. Given
the systemic nature of atherosclerosis, however, the
use of blood pool agents may be beneficial for wholebody screening. A hybrid contrast agent called
gadobenate dimeglumine (MultiHance, Bracco Diagnostics, Milan, Italy) has been approved for use in
Europe and has some protein binding, which has been
shown to improve arterial signal-to-noise significantly
when compared with traditional Gd-chelate contrast at
a comparable Gd dose of 0.1 mmol/kg [111]. Recently,
Ruehm et al [99] demonstrated the feasibility of
performing a five-station bolus-chase MRA using a
0.3-mmol/kg dose of gadobenate dimeglumine.
Any of the aforementioned improvements may
significantly expand the current role and diagnostic
accuracy of aortic and peripheral MRA. Specifically,
they may improve the reliability of infrapopliteal
imaging and even renal imaging during a boluschase MRA. A high percentage of patients with
peripheral vascular disease have renal artery stenosis, yet most of the current bolus-chase techniques
fail to produce diagnostic-quality images of the renal
arteries reliably. Although this may be secondary to
the height of the patient (insufficient superior anatomic coverage of overlapping stations), more commonly time considerations often result in the use of
Fig. 24. Segmented volume bolus chase acquisition (shoot

and scoot). In this acquisition scheme, the low spatial frequency data for each of the three stations are acquired during
the arterial phase of the bolus. Imaging was initiated by an
automated bolus detection algorithm that monitored the
contrast bolus arrival in the mid-abdominal aorta, which
represented the center of the proximal station. After
acquiring the critical central k-space data, the algorithm
returns the table to the proximal stations so that remaining
high-spatial frequency data can be acquired to complete
data acquisition for each station. By segmenting k-space
data acquisition into two separate passes, this technique
shortens the time requirements for the duration of the arterial
enhancement and minimizes the time delay before
imaging of the terminal station. This in turn enables the
use of faster injection rates for improved arterial visualization of the infrapopliteal arteries. (Adapted from Ho VB,
Foo TKF, Czum JM, et al. Contrast-enhanced magnetic
resonance angiography: technical considerations for optimized clinical implementation. Top Magn Reson Imaging
140
and Maureen N. Hood, BSN, RN, RT(MR), of the

Uniformed Services University of the Health Sciences for their invaluable assistance in the preparation
of this manuscript.
References
Fig. 25. Delayed-phase three-dimensional steady-state free

precession MRA. In this 64-year-old man, venous invasion
from a large left inferior pole renal cell carcinoma (white
arrows) is well seen on a three-dimensional steady-state free
precession MRA performed approximately 7 to 8 minutes
following a single dose of Gd-chelate contrast media. This
oblique coronal subvolume maximum intensity projection
clearly depicts the tumor extension from the left renal vein
and into the inferior vena cava (black arrows). (From Ho
VB, Foo TKF, Czum JM, et al. Contrast-enhanced magnetic
resonance angiography: technical considerations for optimized clinical implementation. Top Magn Reson Imaging
thicker partitions (eg, 2.5 to 3 mm thick partitions)

in the aortoiliac station such that renal artery assessment is suboptimal.
Summary
Contrast-enhanced MRA can be an accurate and
reliable method for the arterial evaluation of the
abdominal aorta and peripheral vessels. This technique can be adapted for a variety of anatomic
regions. The basic issues relate to proper synchronization of imaging with peak arterial enhancement
and to optimization of voxel dimensions for adequate
depiction of the arterial structures.
Acknowledgments
The authors thank Michael Schweikert, RT(R),
(MR), lead MR technologist at Doylestown Hospital,
[1] Koelemay MJ, Lijmer JG, Stoker J, et al. Magnetic

resonance angiography for the evaluation of lower
extremity arterial disease: a meta-analysis. JAMA
2001;285:1338 45.
[2] Nelemans PJ, Leiner T, de Vet HC, et al. Peripheral
arterial disease: meta-analysis of the diagnostic performance of MR angiography. Radiology 2000;217:
105 14.
[3] Yucel EK, Anderson CM, Edelman RR, et al. AHA
scientific statement. Magnetic resonance angiography: update on applications for extracranial arteries.
Circulation 1999;100:2284 301.
[4] Axel L. Blood flow effects in magnetic resonance
imaging. AJR Am J Roentgenol 1984;143:1157 66.
[5] Bradley Jr. WG, Waluch V. Blood flow magnetic resonance imaging. Radiology 1985;154:443 50.
[6] Wedeen VJ, Meuli RA, Edelman RR, et al. Projective
imaging of pulsatile flow with magnetic resonance.
Science 1985;230:946 8.
[7] Debatin JF, Spritzer CE, Grist TM, et al. Imaging of
the renal arteries: value of MR angiography. AJR Am
J Roentgenol 1991;157:981 90.
[8] De Haan MW, Kouwenhoven M, Thelissen RP, et al.
Renovascular disease in patients with hypertension:
detection with systolic and diastolic gating in threedimensional, phase contrast MR angiography. Radiology 1996;198:449 56.
[9] Loubeyre P, Trolliet P, Cahen R, et al. MR angiography of renal artery stenosis: value of the combination
of three-dimensional time-of-flight and three-dimensional phase-contrast MR angiography sequences.
[10] Alfidi RJ, Masaryk TJ, Haacke EM, et al. MR angiography of peripheral, carotid, and coronary arteries. AJR Am J Roentgenol 1987;149:1097 109.
[11] Kaufman JA, McCarter D, Geller SC, et al. Two-dimensional time-of-flight MR angiography of the lower extremities: artifacts and pitfalls. AJR Am J
Roentgenol 1998;171:129 35.
[12] Prince MR, Yucel EK, Kaufman JA, et al. Dynamic
gadolinium-enhanced three-dimensional abdominal
MR arteriography. J Magn Reson Imaging 1993;3:
877 81.
[13] Hany TF, Schmidt M, Davis CP, et al. Diagnostic
impact of four postprocessing techniques in evaluating contrast-enhanced three-dimensional MR angiography. AJR Am J Roentgenol 1998;170:907 12.
[14] Udupa JK. Three-dimensional visualization and analysis methodologies: a current perspective. Radiographics 1999;19:783 806.

[15] Ho VB, Foo TK, Czum JM, et al. Contrast-enhanced
magnetic resonance angiography: technical considerations for optimized clinical implementation. Top
Magn Reson Imaging 2001;12:283 99.
[16] Maki JH, Prince MR, Londy FJ, et al. The effects
of time varying intravascular signal intensity and
k-space acquisition order on three-dimensional MR
angiography image quality. J Magn Reson Imaging
1996;6:642 51.
[17] Prince MR. Gadolinium-enhanced MR aortography.
Radiology 1994;191:155 64.
[18] Weiger M, Pruessmann KP, Kassner A, et al. Contrast-enhanced 3D MRA using SENSE. J Magn Reson Imaging 2000;12:671 7.
[19] Earls JP, Rofsky NM, DeCorato DR, et al. Breathhold single dose Gd-enhanced three-dimensional MR
aortography: usefulness of a timing examination and
MR power injector. Radiology 1996;201:705 10.
[20] Hany TF, McKinnon GC, Leung DA, et al. Optimization of contrast timing for breath-hold three-dimensional MR angiography. J Magn Reson Imaging
1997;7:551 6.
[21] Kopka L, Vosshenrich R, Rodenwaldt J, et al. Differences in injection rates on contrast-enhanced breathhold three-dimensional MR angiography. AJR Am J
Roentgenol 1998;170:345 8.
[22] Prince MR, Narasimham DL, Stanley JC, et al.
Breath-hold gadolinium-enhanced MR angiography
of the abdominal aorta and its major branches. Radiology 1995;197:785 92.
[23] Foo TK, Saranathan M, Prince MR, et al. Automated
detection of bolus arrival and initiation of data acquisition in fast, three-dimensional, gadolinium-enhanced
MR angiography. Radiology 1997;203:275 80.
[24] Ho VB, Foo TK. Optimization of gadolinium-enhanced magnetic resonance angiography using an
automated bolus detection algorithm (MR SmartPrep): original investigation. Invest Radiol 1998;33:
515 23.
[25] Schoenberg SO, Bock M, Knopp MV, et al. Renal
arteries: optimization of three-dimensional gadolinium-enhanced MR angiography with bolus-timingindependent fast multiphase acquisition in a single
breath hold. Radiology 1999;211:667 79.
[26] Wilman AH, Riederer SJ, King BF, et al. Fluoroscopically triggered contrast-enhanced three-dimensional
MR angiography with elliptical centric view order:
application to the renal arteries. Radiology 1997;
205:137 46.
[27] Lee VS, Rofsky NM, Krinsky GA. Single-dose
breath-hold gadolinium-enhanced three-dimensional
MR angiography of the renal arteries. Radiology
1999;211:69 78.
[28] Rofsky NM, Johnson G, Adelman MA. Peripheral
vascular disease evaluated with reduced dose gadolinium-enhanced MR angiography. Radiology 1997;
205:163 9.
[29] Boos M, Scheffler K, Haselhorst R, et al. Arterial first
pass gadolinium-CM dynamics as a function of sev-
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
141
eral intravenous saline flush and Gd volumes. J Magn

Schoenberg SO, Essig M, Bock M, et al. Comprehensive MR evaluation of renovascular disease in
five breath holds. J Magn Reson Imaging 1999;10:
347 56.
Johnston KW, Rutherford RB, Tilson MD, et al. Suggested standards for reporting on arterial aneurysms.
J Vasc Surg 1991;13:444 50.
Bengtsson H, Bergqvist D, Sternby N-H. Increasing
prevalence of abdominal aortic aneurysms: a necropsy study. Eur J Surg 1992;158:19 23.
Bengtsson H, Bergqvist D. Ruptured abdominal
aortic aneurysm: a population-based study. J Vasc
Surg 1993;18:74 80.
Ernst CB. Abdominal aortic aneurysm. N Engl J Med
1993;328:1167 72.
Nevitt MP, Ballard DJ, Hallett Jr JW. Prognosis of
abdominal aortic aneurysms. N Engl J Med 1989;
321:1009 14.
The UK Small Aneurysm Trial Participants. Mortality
results for randomized controlled trial of early elective surgery or ultrasonographic surveillance for
small abdominal aortic aneurysms. Lancet 1998;
352:1649 55.
Hilfiker PR, Quick HH, Debatin JF. Plain and covered stent-grafts: in vitro evaluation of characteristics
at three-dimensional MR angiography. Radiology
1999;211:693 7.
Juergen KU, Tombach B, Reimer P, et al. Three-dimensional contrast-enhanced MR angiography of endovascular covered stents in patients with peripheral
arterial occlusive disease. AJR Am J Roentgenol
2001;176:1299 303.
Walsh DW, Ho VB, Haggerty MF. Mycotic aneurysm
of the aorta: MRI and MRA features. J Magn Reson
Imaging 1997;7:312 5.
Nypaver TJ, Shepard AD, Reddy DJ, Elliott Jr. JP,
Smith RF, Ernst CB. Repair of pararenal abdominal
aortic aneurysms: an analysis of operative management. Arch Surg 1993;128:803 13.
Amparo EG, Hoddick WK, Hricak H, Sollitto R, Justich E, Filly RA, et al. Comparison of magnetic resonance imaging and ultrasound in the evaluation of
abdominal aortic aneurysms. Radiology 1985;154:
451 6.
Ruff SJ, Watson MR. Magnetic resonance imaging
versus angiography in the preoperative assessment
of abdominal aortic aneurysms. Am J Surg 1988;
155:651 4.
Valji K. Abdominal aorta. In: Valji K, editor. Vascular and interventional radiology. Philadelphia: WB
Saunders; 1999. p. 80 95.
Kaufman JA, Geller SC, Petersen MJ, et al. MR
imaging (including MR angiography) of abdominal
aortic aneurysms: comparison with conventional
angiography. AJR Am J Roentgenol 1994;163:
203 10.
Laissy JP, Soyer Tebboune D, et al. Abdominal aortic
142
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
[58]
[59]
[60]

aneurysms: assessment with gadolinium-enhanced
time-of-flight coronal MR angiography (MRA). Eur
J Radiol 1995;20:1 8.
Petersen MJ, Cambria RP, Kaufman JA, et al. Magnetic resonance angiography in the preoperative evaluation of abdominal aortic aneurysms. J Vasc Surg
1995;21:891 8.
Prince MR, Narasimham DL, Stanley JC, et al. Gadolinium-enhanced magnetic resonance angiography
of abdominal aortic aneurysms. J Vasc Surg 1995;
21:656 69.
Williams DM, Lee DY, Hamilton BH, Marx MV,
Narasimham DL, Kazanjian SN, et al. The dissected aorta. Part III. Anatomy and radiologic diagnosis of branch-vessel compromise. Radiology
1997;203:37 44.
Ho VB, Prince MR. Thoracic MR aortography: imaging techniques and strategies. Radiographics 1998;18:
287 309.
Ho VB, Prince MR, Dong Q. Magnetic resonance
imaging of the aorta and branch vessels. Coron Artery
Dis 1999;10:141 9.
Prince MR, Narasimham DL, Jacoby WT, Williams
DM, Cho KJ, Marx MV, et al. Three-dimensional
gadolinium-enhanced MR angiography of the
thoracic aorta. AJR Am J Roentgenol 1996;166:
1387 97.
Ruehm SG, Weishaupt D, Debatin JF. Contrast-enhanced MR angiography in patients with aortic occlusion (Leriches syndrome). J Magn Reson Imaging
2000;11:401 10.
Brewster DC, Retana A, Waltman AC, et al. Angiography in the management of aneurysms of the abdominal aorta. Its value and safety. N Engl J Med
1975;292:822 5.
Missouris CG, Buckenham T, Cappuccio FP, et al.
Renal artery stenosis: a common and important problem in patients with peripheral vascular disease. Am J
Med 1994;96:10 4.
Sawicki PT, Kaiser S, Heinemann L, et al. Prevalence
of renal artery stenosis in diabetes mellitus: an autopsy study. J Intern Med 1991;229:489 92.
Caps MT, Perissinotto C, Zierler RE, et al. Prospective study of atherosclerotic disease progression in the
renal artery. Circulation 1998;98:2866 72.
Stewart MT, Smith III RB, Fulenwider JT, et al. Concomitant renal revascularization in patients undergoing aortic surgery. J Vasc Surg 1985;2:400 5.
Tsoukas AI, Hertzer NR, Mascha EJ, et al. Simultaneous aortic replacement and renal artery revascularization: the influence of preoperative renal function
on early risk and late outcome. J Vasc Surg 2001;34:
1041 9.
Knopp MV, Floemer F, Schoenberg SO, et al. Noninvasive assessment of renal artery stenosis: current
concepts and future directions in magnetic resonance
angiography. J Comput Assist Tomogr 1999;23:
S111 7.
Krijnen P, van Jaarsveld BC, Steyerberg EW, et al. A
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
clinical prediction rule for renal artery stenosis. Ann

Intern Med 1998;129:705 11.
Bakker J, Beek FJ, Beutler JJ, et al. Renal artery
stenosis and accessory renal arteries: accuracy of detection and visualization with gadolinium-enhanced
breath-hold MR angiography. Radiology 1998;207:
497 504.
De Cobelli F, Vanzulli A, Sironi S, et al. Renal artery
stenosis: evaluation with breath-hold, three-dimensional, dynamic, gadolinium-enhanced versus threedimensional, phase contrast MR angiography. Radiology 1997;205:689 95.
Hany TF, Leung DA, Pfammatter T, Debatin JF.
Contrast-enhanced magnetic resonance angiography
of the renal arteries: original investigation. Invest
Radiol 1998;33:653 69.
Prince MR, Schoenberg SO, Ward JS, et al. Hemodynamically significant atherosclerotic renal artery
stenosis: MR angiographic features. Radiology
1997;205:128 36.
Snidow JJ, Johnson MS, Harris VJ, et al. Threedimensional gadolinium-enhanced MR angiography
for aortoiliac inflow assessment plus renal artery
screening in a single breath hold. Radiology 1996;
198:725 32.
Steffens JC, Link J, Grassner J, et al. Contrast-enhanced, k-space-centered, breath-hold MR angiography of the renal arteries and the abdominal aorta.
Dong Q, Schoenberg SO, Carlos RC, et al. Diagnosis
of renal vascular disease with MR angiography. Radiographics 1999;19:1535 54.
Hood MN, Ho VB, Corse WR. Three-dimensional
phase-contrast magnetic resonance angiography: a
useful clinical adjunct to gadolinium-enhanced
three-dimensional renal magnetic resonance angiography? Mil Med 2002;167:343 9.
Bass JC, Prince MR, Londy FJ, et al. Effect of gadolinium on phase-contrast MR angiography of the renal arteries. AJR Am J Roentgenol 1997;168:261 6.
Johnson DB, Lerner CA, Prince MR, et al. Gadolinium-enhanced magnetic resonance angiography of renal transplants. Magn Reson Imaging 1997;15: 13 20.
Neimatallah MA, Dong Q, Schoenberg SO, et al.
Magnetic resonance imaging in renal transplantation.
Low RN, Martinex AG, Steinberg SM, et al. Potential
renal transplant donors: evaluation with gadoliniumenhanced MR angiography and MR urography. Radiology 1998;207:165 72.
Rankin SC, Jan W, Koffman CG. Noninvasive
imaging of living related kidney donors: evaluation
with CT angiography and gadolinium-enhanced MR
angiography. AJR Am J Roentgenol 2001;177:
349 55.
Frick MP, Goldberg ME. Uro- and angiographic findings in a normal population; screening of 151
symptom-free potential transplant donors for renal
disease. AJR Am J Roentgenol 1980;134:503 5.

[75] Spring DB, Salvatierra Jr. O, Palubinskas AJ, et al.
Results and significance of angiography in potential
kidney donors. Radiology 1979;133:45 7.
[76] Criqui MH, Fronek A, Barrett-Conner E, et al. The
prevalence of peripheral arterial disease in a defined
population. Circulation 1985;71:510 5.
[77] Helperin JL, Creager MA. Arterial obstructive diseases
of the extremities. In: Loscalzo J, Creager MA, Dzau
VJ, editors. Vascular medicine: a textbook of vascular
biology and diseases. 2nd edition. Boston: Little,
Brown; 1996. p. 825 52.
[78] Adelman MA, Jacobowitz GR. Body MR angiography: a surgeons perspective. Magn Reson Imaging
Clin N Am 1998;6:397 416.
[79] Haimovici H. Patterns of arteriosclerotic lesions of
the lower extremity. Arch Surg 1967;95:918 33.
[80] Imparato AM, Kim GE, Davidson T, et al. Intermittent claudication: its natural course. Surgery 1975;78:
795 9.
[81] Henry M, Amor M, Ethevenot G, et al. Palmaz stent
placement in iliac and femoropopliteal arteries: primary and secondary patency in 310 patients with 2
4-year follow-up. Radiology 1995;197:167 74.
[82] Johnston KW. Iliac arteries: reanalysis of results of
balloon angioplasty. Radiology 1993;186:207 12.
[83] Long AL, Sapoval MR, Beyssen BM, et al. Strecker
stent implantation in iliac arteries: patency and predictive factors for long-term success. Radiology
1995;194:739 44.
[84] Pentecost MJ, Criqui MH, Dorros G, et al. Guidelines
for peripheral percutaneous transluminal angioplasty
of the abdominal aorta and lower extremity vessels.
Circulation 1994;89:511 31.
[85] Dardik H, Kahn M, Dardik I, et al. Influence of failed
vascular bypass procedures on conversion of belowknee to above-knee amputation levels. Surgery 1982;
91:64 9.
[86] Schlenker JD, Wolkoff JS. Major amputations after
femoropopliteal bypass procedures. Am J Surg
1975;129:495 9.
[87] Baum RA, Rutter CM, Sunshine JH, et al. Multicenter trial to evaluate vascular magnetic resonance angiography of the lower extremity. American College
of Radiology Rapid Technology Assessment Group.
JAMA 1995;274:875 80.
[88] Yucel EK, Kaufman JA, Geller SC, et al. Atherosclerotic occlusive disease of the lower extremity: prospective evaluation with two-dimensional time-of-flight
MR angiography. Radiology 1993;187: 637 41.
[89] Carpenter JP, Owen RS, Baum RA, et al. Magnetic
resonance angiography of peripheral runoff vessels.
J Vasc Surg 1992;16:807 13.
[90] Owen RS, Carpenter JP, Baum RA, et al. Magnetic
resonance imaging of angiographically occult runoff
vessels in peripheral arterial occlusive disease. N Engl
J Med 1992;326:1577 81.
[91] Bertschinger K, Cassina PC, Debatin JF, et al. Surveillance of peripheral arterial bypass grafts with
three-dimensional MR angiography: comparison with
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
143
digital subtraction angiography. AJR Am J Roentgenol 2001;176:215 20.

Earls JP, DeSena S, Bluemke DA. Gadolinium-enhanced three-dimensional MR angiography of the entire aorta and iliac arteries with dynamic manual table
translation. Radiology 1998;209:844 9.
Ho KY, Leiner T, de Haan MW, et al. Peripheral
vascular tree stenoses: evaluation with moving-bed
infusion-tracking MR angiography. Radiology 1998;
206:683 92.
Ho VB, Choyke PL, Foo TKF. Automated bolus
chase peripheral MR angiography: initial practical
experiences and future directions of this work-in-progress. J Magn Reson Imaging 1999;10:376 88.
Ho VB, Meaney JFM, Kent KC, et al. Bolus-chase
peripheral MR angiography: technical considerations.
Appl Radiol 2002;31:11 9.
Leiner T, Ho KY, Nelemans PJ, et al. Three-dimensional contrast-enhanced moving-bed infusion-tracking (MoBI-Track) peripheral MR angiography with
flexible choice of imaging parameters for each field
of view. J Magn Reson Imaging 2000;11:368 77.
Meaney JFM, Ridgeway JP, Chakraverty S, et al.
Stepping-table gadolinium-enhanced digital subtraction MR angiography of the aorta and lower extremity arteries: preliminary experience. Radiology 1999;
211:59 67.
Ruehm SG, Hany TF, Pfammatter T, et al. Pelvic
and lower extremity arterial imaging: diagnostic performance of three-dimensional contrast-enhanced
MR angiography. AJR Am J Roentgenol 2000;174:
1127 35.
Ruehm SG, Goyen M, Barkhausen J, et al. Rapid
magnetic resonance angiography for detection of
atherosclerosis. Lancet 2001;357:1086 91.
Schoenberg SO, Londy FJ, Licato P, et al. Multiphase-multistep gadolinium-enhanced MR angiography of the abdominal aorta and runoff vessels. Invest
Radiol 2001;36:283 91.
Shetty AN, Bis KG, Duerinckx AJ, et al. Lower extremity MR angiography: universal retrofitting of
high-field-strength systems with stepping kinematic imaging platforms initial experience. Radiology
2002;222:284 91.
Wang Y, Winchester PA, Khilnani NM, et al. Contrast-enhanced peripheral MR angiography from the
abdominal aorta to the pedal arteries: combined dynamic two-dimensional and bolus-chase three-dimensional acquisitions. Invest Radiol 2001;36:170 7.
Ho KY, de Haan MW, Kessels AG, et al. Peripheral
vascular stenoses: detection with subtracted and nonsubtracted MR angiography. Radiology 1998;206:
673 81.
Leiner T, de Weert TT, Nijenhuis RJ, et al. Need for
background suppression in contrast-enhanced peripheral magnetic resonance angiography. J Magn Reson
Imaging 2001;14:724 33.
Sodickson DK, McKenzie CA, Li W, et al. Contrastenhanced 3D MR angiography with simultaneous ac-
144
quisition of spatial harmonics: a pilot study. Radiology 2000;217:284 9.

[106] Maki JH, Ephron JH, Glickerman DJ, et al. Moving
table Gd-enhanced MR angiography of the lower extremities: a combination 3D and 2D technique preliminary results [abstract]. In: Proceedings of the
Eighth Scientific Meeting and Exhibition of the International Society for Magnetic Resonance in Medicine. Berkeley: International Society for Magnetic
Resonance in Medicine; 2000. p. 1810.
[107] Foo TK, Ho VB, Hood MN, et al. High-spatial-resolution multistation MR imaging of lower-extremity
peripheral vasculature with segmented volume acquisition: feasibility study. Radiology 2001;219:
835 41.
[108] Hany TF, Carroll TJ, Omary RA, et al. Aorta and
runoff vessels: single injection MR angiography with
automated table movement compared with multiinjec-
tion time-resolved MR angiographyinitial results.

Radiology 2001;221:266 72.
[109] Foo TK, Ho VB, Marcos H, et al. MR angiography
using steady state free precession [abstract]. In: Proceedings of the International Society for Magnetic
Resonance in Medicine Ninth Scientific Meeting
and Exhibition and European Society for Magnetic
Resonance in Medicine and Biology Joint Annual
Meeting. Berkeley: International Society for Magnetic Resonance in Medicine; 2001. p. 1950.
[110] Knopp MV, von Tengg-Kobligk H, Floemer F, et al.
Contrast agents for MRA: future directions. J Magn
[111] Volk M, Strotzer M, Lenhart M, et al. Renal timeresolved MR angiography; quantitative comparison
of gadobenate dimeglumine and gadopentetate dimeglumine with different doses. Radiology 2001;220:
484 8.
Radiol Clin N Am 41 (2003) 145 159
MR imaging of the kidneys and adrenal glands

Gary M. Israel, MD*, Glenn A. Krinsky, MD
Division of Abdominal Imaging, Department of Radiology, HW202, New York University Medical Center,
560 First Avenue, New York, NY 10016, USA
Advances in MR imaging technology, including

high-performance gradients, faster pulse sequences,
and phased-array coils, provide for the acquisition of
near isotropic three-dimensional data sets during a
single breath-hold. Volumetric rendering of these data
provides surgically relevant information for minimally invasive renal-adrenal surgery and allows
MR imaging to compete with multidetector CT.
Further advantages of MR imaging include lack of
ionizing radiation, direct multiplanar capability that
enables more accurate localization of masses, and
superior intrinsic soft tissue contrast augmented by
the use of extracellular gadolinium chelates. Gadolinium chelates have been shown to be exceedingly
safe, may be administered to patients without concern
for contrast-induced nephrotoxicity, and are well
tolerated in those patients with a history of iodinated
contrast allergy [1,2]. Finally, the ability of MR
imaging to detect both gross and microscopic fat
provides for accurate characterization of adrenal and
renal masses.
Technique
With the evolution of MR imaging technology, the
protocols used to evaluate the kidneys and adrenal
glands have also evolved. At the authors institution,
all abdominal MR imaging examinations are performed with a torso phased-array coil. Phased-array
coils increase signal to noise by a factor of two to
three, which allows for the use of smaller fields of
view with concomitant increased spatial resolution.
E-mail address: gary.israel@med.nyu.edu (G.M. Israel).
Breath-hold sequences are used exclusively to minimize artifacts secondary to respiratory motion. Studies are performed during end expiration to optimize
image co-registration for subtraction algorithms. For
those patients in whom the sequences are longer than
their breath-hold capability, the authors administer
2 L/minute of oxygen by nasal cannula. Finally,
before starting the MR imaging examination, cushions are used to elevate the patients arms anterior to
the level of the kidneys and out of the imaging plane
of a coronal acquisition. This minimizes wraparound
artifact (in the phase-encoding direction) when performing three-dimensional coronal acquisitions and
allows the use of smaller fields of view with improved resolution.
Comprehensive examination of the kidneys entails
evaluating the renal vasculature, parenchyma, and
collecting system. Precontrast imaging includes an
axial breath-hold T1-weighted gradient echo (GRE)
sequence performed in and out of phase. This
sequence provides an excellent anatomic overview
of the abdomen and is useful to evaluate for adenopathy and characterize an incidental adrenal lesion.
Used in conjunction with a frequency-selective fatsuppressed T1-weighted sequence, this also allows
differentiation of fat from hemorrhage, both of which
may occur in a renal or adrenal mass.
A coronal breath-hold T2-weighted half-Fourier
single-shot turbo spin echo sequence is performed to
help characterize cystic lesions of the kidney, to
assess for hydronephrosis, and to determine if an
adrenal mass is present. The coronal plane is advantageous in evaluating exophytic lesions that occur at
the poles of the kidneys. These may not be optimally
demonstrated in the axial plane. In addition, coronal
plane images are more helpful in establishing the
association of the lesion to its surrounding organs.
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 5 9 - 3
146
G.M. Israel, G.A. Krinsky / Radiol Clin N Am 41 (2003) 145159
To evaluate the renal vasculature, a high-resolution breath-hold fat-suppressed three-dimensional

T1-weighted spoiled GRE sequence is performed in
a coronal-oblique plane before and after the intravenous administration of gadolinium. Using the
proper scan delay and acquiring the low spatial
frequency (high contrast) lines of k-space during
peak arterial enhancement are critical to minimize
venous contamination and to prevent scanning before
sufficient contrast reaches the aorta. Optimization of
the arterial phase may be determined by using
fluoroscopic triggering [3], an automated bolus
detection technique [4], a timing run, or with a best
guess method. As gradient strength and pulse
sequences improve, a time-resolved approach can
be used, which obviates the need for a timing
strategy [5]. By scanning with rapid temporal resolution (approximately 5 seconds) at least one phase
shows optimal arterial opacification.
The authors prefer a timing run using 1 mL of
gadolinium followed by a 20-mL saline flush injected
at 2 mL/second by a power injector [6]. The MR
angiogram is then performed with 19 mL of gadolinium followed by a 20-mL saline flush. Approximately 30 seconds after the angiogram, the same
sequence is repeated to obtain an MR venogram.
With the near isotropic resolution of a threedimensional sequence, it is possible to evaluate and
reformat the images in innumerable planes using a
workstation. It is also possible to display the data as
maximum-intensity-projection (MIP) images, which
have a similar appearance to conventional angiography. With the MIP algorithm, however, small vessels
may not be depicted and stenoses may be overestimated. It is always necessary to review the
three-dimensional source data to confirm the findings
of the MIP images.
Evaluation of the renal parenchyma is performed
with a second breath-hold three-dimensional fat-suppressed T1-weighted spoiled GRE sequence in the
axial plane [7]. This is performed before and after
the administration of intravenous gadolinium. The
postcontrast acquisition is obtained after the MR
venogram, approximately 3 to 5 minutes after the
gadolinium bolus. This is the most important
sequence the authors use in characterizing a renal
mass as an enhancing neoplasm or a cyst. In the
authors experience, some cystic neoplasms are so
hypovascular that at least 2 minutes are needed to
demonstrate enhancement. In many cases, it is possible to determine qualitative enhancement of a lesion
with side-by-side comparison of the precontrast and
postcontrast acquisitions. In those cases in which a
lesion is hyperintense on the precontrast images,
however, qualitative enhancement is difficult, if not

impossible, to appreciate. In this instance, a subtraction algorithm may be applied to help assess
enhancement characteristics (Fig. 1) [8]. With good
image co-registration, subtracted images appear similar to fat-suppressed postcontrast images. In those
cases of poor image co-registration, however, a
careful evaluation of the nonsubtracted images
is necessary.
MR urography may be performed with T2weighted turbo spin echo sequences using a thick
slab projection technique or with multiple contiguous
thin sections. Alternatively, a delayed three-dimensional T1-weighted GRE sequence after the administration of gadolinium contrast material can be
performed [9]. The authors prefer the latter technique
because the spatial resolution is much higher and
the voxels are near isotropic, which provides for
excellent image quality when viewed at any projection. Immediately after the timing run, all patients
receive 10 mg of intravenous furosemide to augment
diuresis. After the parenchymal axial three-dimensional T1-weighted GRE sequence, the vascular
coronal-oblique three-dimensional T1-weighted sequence is repeated a final time to evaluate the collecting system and ureter (approximately 12 minutes
after the initial injection of contrast). The precontrast
acquisition is then subtracted from the delayed MR
urographic acquisition. A MIP image is created,
which has a similar appearance to a conventional
urogram (Fig. 2).
Renal mass characterization

With the exception of angiomyolipoma, routine
MR sequences have not been shown to be sensitive
or specific in the characterization of renal masses.
The most important aspect in analyzing a renal mass
is to demonstrate the presence or absence of
enhancement. An enhancing mass implies a vascular
mass, consistent with a neoplasm. Once a lesion has
been shown to demonstrate enhancement, it is necessary to characterize it as a surgical lesion (renal cell
carcinoma, oncocytoma, or transitional cell carcinoma) or a nonsurgical lesion (metastases, lymphoma, or angiomyolipoma).
Renal cell carcinoma
Renal cell carcinoma is the most common renal
neoplasm accounting for 80% to 85% of all malignant renal tumors and for 2% of all cancers [10]. The
147
Fig. 1. (A) Unenhanced sagittal fat-suppressed T1-weighted GRE image shows a 2-cm hyperintense mass in the upper pole of the
kidney (arrow). (B) Nephrogenic phase of enhancement shows the mass (arrow) is now isointense to the renal parenchyma. It is
difficult to determine subjectively the presence or absence of enhancement. (C) An image obtained by subtracting the precontrast
(A) from the postcontrast image (B) shows the mass (arrow) as markedly hypointense, without internal enhancement, consistent
with a hemorrhagic cyst. Also notice the simple cyst (c) anteriorly and the enhancing renal neoplasm (N ) at the posterior inferior
aspect of the kidney.
widespread use of cross-sectional imaging and the

incidental detection of asymptomatic neoplasms have
increased the incidence of renal cell carcinoma [10].
Combined with the improved characterization of
small renal lesions and earlier surgical intervention,
there has been a slight improvement in the 5-year
survival of renal cell carcinoma [10].
Renal cell carcinomas have variable signal intensity on T1- and T2-weighted sequences. With regard to
the background renal parenchyma, they are often
slightly hypointense on T1-weighted images and isointense to slightly hyperintense on the T2-weighted
images. Renal cell carcinomas, however, may dem-
onstrate any signal intensity depending on their content of hemorrhagic material. The diagnosis of renal
cell carcinoma rests on demonstrating enhancement
within a renal mass. Unlike CT, in which Hounsfield
units are standardized, MR intensity units are arbitrary
and vary from sequence to sequence. It is difficult
to accurately quantify enhancement. Qualitative
enhancement by means of a subjective comparison
of the precontrast and postcontrast acquisitions or
more accurately by means of subtraction algorithms
needs to be performed. This is most important for
hypovascular lesions or those that are hyperintense on
the precontrast T1-weighted images.
148
patients are at increased risk. In these patients with

reduced numbers of nephrons, this can be performed
safely with MR imaging without exposure to nephrotoxic contrast agents.
Angiomyolipoma
Fig. 2. Maximum-intensity-projection of a delayed coronal

gadolinium-enhanced three-dimensional fat-suppressed
T1-weighted GRE acquisition shows contrast material in
the renal collecting systems (arrow) and ureters (arrows)
resembling a conventional urogram.
Angiomyolipoma (renal hamartoma), a benign

tumor, is composed of varying amounts of fat,
smooth muscle, and blood vessels. They are uncommon lesions with a prevalence of 0.3% to 3% and
occur more commonly in women than men [12].
Angiomyolipomas occur in two different clinical
scenarios. More commonly, they are sporadic
(80%); however, they may be associated with tuberous sclerosis (20%) in which they tend to be multiple
and bilateral [13]. Patients are usually asymptomatic
and angiomyolipomas are usually incidentally discovered when the patient is imaged for another
reason. When large, angiomyolipomas may exert
mass effect on the adjacent organs and cause symptoms. In addition, patients with large angiomyolipomas may present with acute flank pain caused by
spontaneous hemorrhage. This may be life threatening and require emergent laparotomy.
The prognosis of renal cell carcinoma is related to

the tumor stage. MR imaging has been shown to be
accurate for the staging of renal cell carcinoma and
more accurate than CT for the evaluation of tumor
extension into the renal vein and inferior vena cava
[11]. It is important to demonstrate the most cephalad
extent of thrombus in the inferior vena cava, because
the surgical approach is altered if thrombus approaches the right atrium. For this the multiplanar
capability of MR imaging is ideally suited. In addition, it is often possible to differentiate enhancing
tumor thrombus from bland thrombus (Fig. 3).
MR imaging does not offer any advantage when
compared with CT for evaluating retroperitoneal
adenopathy. The signal characteristics of metastatic
lymph nodes are similar to those of normal lymph
nodes. As with CT, evaluation for metastatic adenopathy is size dependent, with lymph nodes greater
than 1 cm considered abnormal and suspicious for
metastatic disease.
After surgery, gadolinium-enhanced MR imaging
may be used to evaluate for early postoperative
complications including hemorrhage or urinary leak
in those patients who undergo partial nephrectomy
(Fig. 4). In addition, MR imaging is useful in the
routine postoperative surveillance for recurrent neoplasm or metachronous lesion for which these
Fig. 3. Coronal reformation from an axial gadoliniumenhanced fat-suppressed three-dimensional T1-weighted

GRE acquisition shows a large complex enhancing left
renal mass (long straight arrows) consistent with a renal cell
carcinoma. Enhancing tumor thrombus extends into the left
renal vein (curved arrows) and inferior vena cava (short
straight arrows) to just below the right hemidiaphragm.
Notice the nonenhancing bland thrombus (T ) in the infrarenal inferior vena cava.
149
Fig. 4. (A) Coronal gadolinium-enhanced fat-suppressed T1-weighted image obtained during the excretory phase demonstrates a
complex enhancing mass (straight arrow), which abuts the lower pole calyx (curved arrow). (B) Following partial nephrectomy,
maximum-intensity-projection image from a coronal-subtracted gadolinium-enhanced MR urogram demonstrates a urinary leak
(arrows) into the postoperative bed.
Angiomyolipoma is the only renal tumor that may

be characterized on the basis of its tissue composition
and signal characteristics. The relative amounts of fat,
smooth muscle, and vessels within the tumor establish
its MR imaging appearance. The diagnosis of angiomyolipoma rests on demonstrating the presence of
macroscopic fat within the lesion. When an angiomyolipoma is predominately composed of fatty tissue, it
demonstrates hyperintense signal on the T1-weighted
images. Other renal masses, however, including
hemorrhagic cysts may also show similar signal characteristics. It is imperative to compare the T1-weighted
images obtained with frequency-selective fat-suppression with those obtained without fat-suppression, to
establish the presence or absence of macroscopic fat
(Fig. 5). The use of frequency-selective fat-suppression is essential, because hemorrhage and other tissues
with a short T1 lose signal on inversion recovery pulse
sequences and may be diagnosed erroneously as containing fat. Some angiomyolipomas contain only a tiny
amount of macroscopic fat and a concerted effort
should be made to identify even small amounts of
fat. In rare instances these lesions may not contain any
fat. In such cases, the diagnosis of angiomyolipoma
cannot be made and the lesion is indistinguishable
from a renal cell carcinoma.
Angiomyolipoma may also be diagnosed with the
use of chemical-shift imaging techniques. Exploiting
the precessional frequency differences of fat and
water, this technique provides images when fat and
water signal are in phase (additive) or out of phase
(destructive). This produces the characteristic India
ink artifact on the T1-weighted out-of-phase images,
manifested as a low signal intensity rim at any soft
tissue (water) and fat interface. Both hemorrhagic
cysts and angiomyolipomas are hyperintense on
150
Fig. 5. (A) Axial T1-weighted GRE image (in phase) shows a 1.5-cm hyperintense left renal mass (arrow). This is nonspecific,
and may represent a hemorrhagic cyst, angiomyolipoma, or, less likely, a renal neoplasm. (B) Axial T1-weighted GRE image
obtained with frequency-selective fat suppression demonstrates near complete signal loss, diagnostic of an angiomyolipoma.
T1-weighted in-phase images and may be indistinguishable from each other. They are readily differentiated, however, on the T1-weighted out-of-phase
images. For angiomyolipomas, the India ink artifact
appears at the interface of the tumor (fat) with the
kidney (water) (Fig. 6). For hemorrhagic cysts, the
India ink artifact occurs at the interface of the cyst
(fluid) and the perirenal fat (fat) (Fig. 7), not at the
interface of the cyst and the kidney.
Caution should be used in diagnosing a renal mass
as an angiomyolipoma if it loses signal on out-ofphase imaging, because clear cell carcinoma of the
kidney may show identical findings [14]. Clear cell
carcinoma does not contain bulk fat, however, and
does not lose signal on frequency-selective fat-suppressed T1-weighted images.
Metastases
The most common tumor to metastasize to the
kidney is carcinoma of the lung. Renal metastases
tend to be multiple and bilateral, and frequently are
associated with metastases to other organs. Although
they have nonspecific MR imaging features, renal
metastases may demonstrate infiltrative growth patterns. With the proper clinical history, the diagnosis
Lymphoma
Lymphoma may involve the kidneys by hematogenous spread, in which a single mass or multiple
bilateral masses are present, or by direct extension of
retroperitoneal lymphoma. Generally, most patients
with renal lymphoma have systemic involvement and
the diagnosis should not be difficult, given the
appropriate clinical history. The MR imaging appearance of lymphoma is nonspecific; however, the most
common appearance is that of multiple homogeneous
solid masses that may be well defined, but tend to
have infiltrative margins with the kidney. When
lymphoma diffusely infiltrates a kidney, the kidney
enlarges, but maintains its reniform shape [15].
Fig. 6. Opposed-phase axial T1-weighted GRE image of the

same angiomyolipoma depicted in Fig. 5 demonstrates the
India ink artifact (arrows) at the interface of the kidney and
the mass, diagnostic of a fat-containing lesion.
151
Fig. 7. (A) Axial T1-weighted GRE image (in phase) demonstrates a hyperintense mass (arrow) in the left kidney. (B) Opposedphase axial T1-weighted GRE image demonstrates the India ink artifact (arrows) at the interface of the mass and the perirenal fat.
Although this may represent a hemorrhagic cyst or neoplasm, this excludes an angiomyolipoma. Subtracted images (not shown)
did not demonstrate enhancement in this lesion, consistent with a hemorrhagic cyst.
should be obvious. In a patient with a history of

malignancy (without other metastases) and a solitary
renal mass, however, the renal mass is more likely to
represent a renal cell carcinoma, and not a metastasis
[16]. Nevertheless, it is possible that a single renal
metastasis could occur, and differentiation from a
renal cell carcinoma may not be obvious. In this
situation, a renal biopsy is indicated to determine the
exact etiology of the lesion.
Simple and complex renal cysts
The appearance of complex cystic renal masses is
diverse and the proper management of these lesions is
frequently not clear-cut. Bosniak [17 20] has proposed a classification system designed to help categorize cystic lesions into surgical and nonsurgical
cases. Although the classification scheme is based on
CT criteria, the same approach provides a useful
framework for MR imaging [21]. It should be
stressed, however, that there is not always a clear
correlation between the findings at MR imaging and
the CT images, and further work is needed to identify
these differences. For example, just as septations may
be identified more readily with ultrasound than with
CT, the same may be true for MR imaging.
When evaluating a complex cystic renal mass on
an MR examination, it is necessary to analyze the
various components of the lesion. This includes the
number of septae, the thickness of the wall or septae,
the interface of the lesion with the kidney, the contents

of the lesion, and most importantly, the presence or
absence of enhancing soft tissue components.
Simple cysts (category I) are common and demonstrate hypointense signal on T1-weighted images
and are uniformly hyperintense on T2-weighted
images. After gadolinium, they do not enhance.
Category II cysts are benign and mildly complex.
They may contain very thin septae that are best
depicted on the T2-weighted images, where they
appear as thin low signal intensity curvilinear structures against the hyperintense cystic fluid (Fig. 8).
When these lesions contain hemorrhagic or proteinaceous material, they demonstrate hyperintense signal
on the T1-weighted images.
MR imaging is ideally suited for characterizing
hemorrhagic cysts, particularly in those patients who
cannot receive iodinated contrast secondary to a
history of renal failure or allergy. This is especially
true in patients with acquired cystic disease of dialysis or with autosomal-dominant polycystic kidney
disease, in which hemorrhagic cysts are very common. Using a subtraction algorithm, it is possible to
demonstrate that these lesions do not enhance, and
thereby characterize them as benign hemorrhagic
cysts (see Fig. 1).
Category III lesions are more complex. They
may demonstrate thick enhancing walls or thick
enhancing septae, but do not contain nodular enhancing soft tissue components associated with the wall
152
because the calcification would not be depicted on

the MR imaging examination, and any possible
enhancement could be appreciated better [22].
Role of MR imaging in preoperative planning
Fig. 8. Coronal T2-weighted half-Fourier single-shot turbo

spin echo image demonstrates a cystic mass in the left
kidney, which contains a thin septation (arrow) consistent
with category II cyst.
or septae (Fig. 9). These lesions are indeterminate

and in most cases surgery is indicated. Approximately 50% of these lesions are malignant [19].
Category IV lesions are those cystic masses that
are clearly malignant and demonstrate unequivocal
enhancing soft tissue components.
Differentiation of a more complex category II cyst
from a less complicated category III cyst is where the
most disagreement occurs among radiologists. This
differentiation is critical, however, because category
II lesions are benign and do not require treatment,
whereas category III lesions, in most cases, require
surgery. A subcategory, category IIF, was proposed.
Category IIF cysts can be followed-up with additional
examinations, and if there is any interval growth, the
lesion has to be considered as category III or higher
and needs surgical evaluation. Without exposure to
radiation or nephrotoxic contrast material, MR
imaging is ideal for following these lesions.
A limitation of MR imaging in characterizing
cystic renal masses is the inability to depict calcification within the wall or septum of a lesion. With CT, it
is sometimes difficult to determine enhancement of a
heavily calcified lesion. Theoretically, MR imaging
would be helpful in characterizing these lesions
With the recent advances in minimally invasive

surgical techniques and improved characterization of
small renal masses, many patients with renal cell
carcinoma are eligible for laparoscopic nephrectomy
or nephron-sparing surgery. By performing a comprehensive renal MR imaging examination with
three-dimensional sequences, it is not only possible
to stage the tumor, but also to demonstrate accurately
the vascular supply and the relationship of the tumor
to the collecting system and the surrounding renal
parenchyma. This helps the surgeon decide which
treatment option is most appropriate and helps minimize any potential complications. With the limited
field of view of the laparoscope, this information is
especially valuable to the surgeon. The authors
encourage their urologists to view and interact with
the three-dimensional data sets at a workstation
before surgery.
MR imaging of the adrenal glands

Similar to the increased detection of asymptomatic renal masses, the widespread use of crosssectional imaging has also increased the detection
Fig. 9. Axial gadolinium-enhanced fat-suppressed threedimensional T1-weighted GRE image demonstrates a 3.5-cm
mass in the right posterior aspect of a horseshoe kidney. The
mass has a thick enhancing wall (straight arrow) and septum
(curved arrow) consistent with a category III cyst. At
pathology, this represented a renal cell carcinoma.
of incidental adrenal masses. Benign and malignant

lesions of the adrenal glands are common and
characterization of these lesions is of great clinical importance.
With MR imaging, it is possible to characterize
some adrenal lesions by means of their signal characteristics on different pulse sequences or by their
enhancement characteristics. These include adenoma,
myelolipoma, hematomas, and cysts. In addition, the
multiplanar capability of MR imaging allows
improved depiction of the relationship of the adrenal
gland to the kidney. It is sometimes difficult to differentiate an exophytic lesion arising from the upper pole
of the kidney from an adrenal lesion, a relationship not
optimally demonstrated with conventional axial
images. The capability of obtaining images in innumerable planes is an advantage of MR imaging.
Adrenal adenoma
The incidence of adrenal adenomas in the general
population is estimated at 2% to 8% [23]. The adrenal
gland is also the most common site of metastases per
unit weight of any organ [24]. Within the oncologic
population, it is common to find an adrenal mass, and a
frequent clinical problem is to determine the etiology
of such a lesion. MR imaging accurately can distinguish an adenoma from a metastasis in most cases.
This allows for more accurate staging of cancer
patients, decreases the number of adrenal biopsies,
and allows the appropriate treatment regimen to be
instituted earlier.
Various MR techniques have been proposed to
characterize adrenal masses as adenoma or a metastasis. Early work demonstrated that calculating T2
153
values could characterize these lesions [25,26]. Sufficient overlap exists, however, to render this cumbersome technique unreliable. Some authors advocate
using gadolinium chelates to help characterize adenomas. Krestin et al [27] showed that adenomas tend to
washout faster when compared with metastasis.
Although some overlap between benign and malignant lesions occurs, it is time consuming to quantify
enhancement on MR, and can be done quickly
with CT. Other authors have reported that adenomas
have a capillary blush seen on arterial-phase imaging,
whereas metastases do not [28]. The easiest, fastest,
and most reliable way to diagnosis an adrenal adenoma, however, rests on demonstrating intracellular
lipid within the mass (lipid-rich adenoma) [29].
By using chemical-shift techniques (breath-hold T1weighted GRE images in phase and out of phase) it is
possible to characterize lipid-rich adenomas. These
adenomas contain intracellular lipid and water protons within the same imaging voxel. On out-of-phase
images, the signal from these protons cancel each
other out and result in signal loss when compared
with the in-phase images (Fig. 10).
Frequently, signal loss on opposed-phase imaging
is obvious. There are cases, however, in which the
signal loss is subtle and not readily apparent. In these
cases, it is necessary to compare the adrenal mass
with an internal standard. In general, the liver is not a
reliable standard secondary to the possibility of
coexisting steatosis. The authors prefer to use spleen
as an internal standard for subjective analysis of
signal loss [29,30]. It is important to remember that
the echo time of the out-of-phase image should be
shorter than the in-phase image to eliminate T2 decay
as a confounding variable of signal loss.
Fig. 10. (A) Axial T1-weighted (in phase) GRE image demonstrates a 2.5-cm right adrenal mass (arrow), which is isointense in
signal when compared with the spleen. (B) Opposed-phase axial T1-weighted GRE image shows the adrenal mass (arrow) is
now hypointense in signal when compared with the spleen, diagnostic of an adrenal adenoma.
154
It is important to carefully evaluate the entire

adrenal mass for signal loss on opposed-phase
imaging. This is especially true for patients with a
known neoplasm that has a high pretest probability to
metastasize to the adrenal glands (lung cancer). A
collision tumor results when a metastasis and an
adenoma are contiguous [31]. In this instance, the
adenomatous portion of the lesion, which contains
intracellular fat, loses signal on out-of-phase images,
whereas the metastatic (nonadenomatous) portion
does not (Fig. 11). A much more common scenario
is coexistence of lipid-rich and lipid-poor regions
within the same adenoma. Under these circumstances
the lesion cannot be characterized definitively as
benign and may require further imaging.
Adrenal adenomas that do not contain intracellular
lipid (lipid-poor adenomas) do not lose signal on
opposed-phased imaging. Furthermore, they cannot
be characterized with CT densitometry (they measure
greater than 17 HU) [32]. These lesions are indeterminate, and are especially troublesome in the
oncologic patient because metastasis cannot be
excluded. It has been demonstrated that these lipidpoor adenomas can be characterized by means of
their washout characteristics on contrast-enhanced
CT or MR imaging [33 35]. Alternatively, positron
emission tomography scanning accurately can characterize lipid-poor adenomas from metastasis [36,37].
Adrenal adenomas may be classified as hypersecreting or, more commonly, as nonhypersecreting.
Hypersecreting adrenal adenomas may produce

aldosterone (Conns syndrome), cortisol (Cushings
syndrome), or androgens (hyperandrogenism).
Because intracellular lipid may be present in both
hypersecreting and nonhypersecreting adenomas, it is
not possible to differentiate them with MR imaging
alone and correlation with the appropriate laboratory
values is necessary.
Adrenal cortical carcinoma
Adrenal cortical carcinoma is a rare neoplasm that
most commonly occurs in the fourth to fifth decades
of life with equal prevalence in men and women.
They are typically large (> 5 cm) at presentation, may
contain varying degrees of hemorrhage and necrosis,
and often contain calcium [23,38]. Some adrenal
cortical carcinomas are hypersecreting and present
earlier and at a smaller size when compared with
nonhypersecreting tumors. The most common hormone produced is cortisol, which manifests as Cushings syndrome.
The signal intensity of adrenal cortical carcinomas
is variable and they generally are heterogeneous on
T1- and T2-weighted sequences. This is secondary to
necrosis and hemorrhagic components that are common within these lesions. After the administration of
gadolinium, the viable portion of the tumor enhances.
Because this neoplasm originates from the adrenal
cortex, it is not surprising that intracellular lipid may
Fig. 11. Two left adrenal masses in a patient with a history of lung cancer. (A) Axial T1-weighted (in phase) GRE image
demonstrates two masses (arrows) in the left adrenal gland, both of which are isointense in signal when compared with the
spleen. (B) Opposed-phase axial T1-weighted GRE image demonstrates that the posterior mass (curved arrow) is slightly
hypointense in signal when compared with the spleen, consistent with an adenoma. The anterior mass (straight arrow) remains
isointense to the spleen, and may represent a lipid-poor adenoma or, in this case, a metastasis.
be present in a portion of the mass. It is possible

that some of the lesion loses signal on the out-of-phase
T1-weighted images, similar to an adenoma [39].
In most cases, however, differentiation from an adenoma should not be difficult. For an adenoma, the
entire lesion should lose signal on the out-of-phase
images as compared with an adrenal carcinoma, in
which only a portion of the lesion drops out in signal.
In addition, adrenal cortical carcinomas are typically
larger (>5 cm) than adenomas; are frequently necrotic;
and may be poorly marginated. In those cases in which
radiologic differentiation is difficult, however, laparoscopic adrenalectomy can be performed [40].
Adrenal cortical carcinomas may grow to be very
large and may directly invade adjacent organs including the kidney, liver, spleen, pancreas, and diaphragm. At times, it may be difficult to determine
the exact organ of origin, especially when a normal
adrenal gland cannot be identified. Imaging in the
coronal or sagittal plane is very helpful in showing
the relationship of the tumor to its surrounding
structures and demonstrating the suprarenal location
of an adrenal mass. Furthermore, adrenal cortical
155
carcinoma has a predilection to invade the adrenal

veins, grow into the renal vein and inferior vena cava,
and extend cephalad toward the heart. Gadoliniumenhanced MR imaging can clearly demonstrate the
venous extension of an adrenal cortical carcinoma
(Fig. 12). It is critical to include a pheochromocytoma
in the differential diagnosis of cortical carcinoma,
because their imaging features may be identical and
failure to do so may result in catastrophic consequences in the operating room.
Pheochromocytoma
Pheochromocytomas are neoplasms of the adrenal
medulla that produce catecholamines. They occur
with equal frequency in men and women, and most
commonly occur during the third and fourth decades
of life. Pheochromocytomas are extra-adrenal, bilateral, or malignant 10% of the time. Although most
commonly sporadic, pheochromocytomas may be
associated with other syndromes including multiple
endocrine neoplasia, von Hippel-Lindau disease, and
neurofibromatosis [41]. Although patients may be
Fig. 12. (A) Sagittal gadolinium-enhanced fat-suppressed T1-weighted GRE image demonstrates a large heterogeneous
enhancing adrenal neoplasm (arrows) displacing the kidney inferiorly. Note the simple cyst (c) in the kidney. (B) Coronal
gadolinium-enhanced fat-suppressed T1-weighted GRE image shows enhancing tumor thrombus growing into the left adrenal
vein (long straight arrow) and inferior vena cava (short straight arrow) with extension into the right atrium (long curved arrow).
Note the bland thrombus proximal to the tumor thrombus (short curved arrow).
156
symptomatic, the symptoms are nonspecific, and

include palpitations, headache, sweating, and hypertension [42]. Even though hypertension is one of the
more common presentations, pheochromocytomas
account for the cause of hypertension in less than
1% of patients [42,43].
The MR imaging appearance of pheochromocytoma has classically been described as markedly
hyperintense on T2-weighted sequences [44,45]. Subsequently, it has been demonstrated that pheochromocytomas may have variable signal on T2-weighted
sequences, especially when they are greater than 5 cm
[45]. MR imaging is more useful in identifying an
adrenal mass in a patient who is clinically thought to
have a pheochromocytoma than in characterizing an
adrenal mass as a pheochromocytoma. Furthermore,
MR imaging is useful in identifying extra-adrenal
pheochromocytomas (paragangliomas) in the retroperitoneum along the paraspinal muscles. Confirmation with nuclear medicine studies is useful in
equivocal cases.
Myelolipoma
Adrenal myelolipoma is a rare nonfunctioning
benign neoplasm that contains a variable amount of
hematopoietic tissue and fat. Calcification can be
seen in approximately 20% of cases [46]. In general,
they are asymptomatic and are incidental findings at
ultrasound, CT, or MR imaging, but may cause pain
if they hemorrhage or are large enough to exert mass
effect on the adjacent organs.
The diagnosis of myelolipoma rests on the demonstration of macroscopic fat within an adrenal mass.
With MR imaging, the fatty portion of the lesion is
hyperintense on T1-weighted images. This is nonspecific and can be seen in any lesion that contains
hemorrhage. Just as in diagnosing a renal angiomyolipoma, it is necessary to perform a frequency-selective fat-suppressed T1-weighted sequence and
compare it with the non fat-suppressed T1-weighted
sequence. The fatty portion of the lesion should lose
signal on the fat-suppressed sequence, and is diagnostic of a myelolipoma (Fig. 13). Myelolipomas
may also be diagnosed with chemical-shift imaging by
identifying the India ink artifact at the interface of the
bulk fat and soft tissue components of the lesion.
When a predominately fatty adrenal myelolipoma
becomes large and exerts mass effect on the adjacent
organs, it may become difficult to ascertain that it
arises from the adrenal gland. In this instance, a
myelolipoma may be confused with a well-differentiated retroperitoneal liposarcoma or even an exophytic renal angiomyolipoma. By obtaining images in
the coronal or sagittal planes, however, it is usually

possible to demonstrate that the lesion has a smooth
interface with the kidney and that there is no defect in
the renal cortex. This finding is highly suggestive that
the lesion does not arise from the kidney, excluding
an angiomyolipoma. Furthermore, a liposarcoma is
expected to engulf or displace the adrenal gland. If a
normal adrenal gland is identified, a myelolipoma
may be excluded. An adrenal gland that contains a
large myelolipoma is expected to be stretched around
the periphery of the tumor, or if the tumor is large
enough, not be seen at all.
Adrenal cysts, pseudocysts, and hematomas
Cysts and pseudocysts of the adrenal gland are rare
and are usually incidentally discovered on cross-sectional imaging. Patients with these lesions are usually
asymptomatic unless they are large enough to produce
mass effect on adjacent organs (Fig. 14). Adrenal cysts
have been subdivided into four main categories:
(1) endothelial (angiomatous or lymphangiectatic);
(2) epithelial; (3) pseudocysts; and (4) parasitic. Pseudocysts may be posttraumatic or postinfectious.
At MR imaging, simple adrenal cysts are usually
hypointense on T1-weighted and hyperintense on
Fig. 13. Axial T1-weighted GRE image demonstrates a 1-cm

hyperintense right adrenal mass (arrow), which may
represent a myelolipoma or a hemorrhagic lesion. Fatsuppressed T1-weighted image (not shown) demonstrated
complete loss of signal, diagnostic of a myelolipoma.
157
Fig. 14. Coronal T2-weighted (HASTE) image demonstrates a large simple cystic suprarenal mass (c) consistent with a
congenital adrenal cyst. The adrenal gland is stretched along the periphery of the mass (straight arrow). Note how the cyst
deviates the left kidney (curved arrow) inferiorly, and exerts mass effect on the pancreas ( P) and spleen (S ).
T2-weighted images. Some pseudocysts may contain

hemorrhage, however, and their signal intensity on
different pulse sequences can vary. The wall of an
adrenal cyst should be thin, without nodular or
enhancing components. In addition, calcification
may occur within the wall, which is depicted better
with CT than with MR imaging. Use of a long echotime (>15 milliseconds) and short flip angle gradient-echo sequence, however, usually can identify
susceptibility artifact from calcium or hemosiderin.
If only unenhanced MR images are obtained it
may be difficult, if not impossible, to differentiate an
adrenal cyst or pseudocyst from a classic pheochromocytoma, both of which are hyperintense on
T2-weighted sequences. In this situation, a gadolinium-enhanced study can reliably distinguish these
two entities because a pheochromocytoma enhances
and an adrenal cyst or pseudocyst does not.
Summary
By performing a comprehensive MR imaging
examination, it is not only possible accurately to
characterize and stage cystic and solid lesions of the
kidney, but also to provide important preoperative
information to the surgeon. In addition, MR imaging
can characterize many adrenal lesions and frequently
can obviate the need to obtain biopsies. The continued development and growth of MR technology
combined with the current trend toward minimally

invasive surgery will expand the role of MR imaging
in the future.
References
[1] Nelson KL, Gifford LM, Lauber-Huber C, et al. Clinical safety of gadopentetate dimeglumine. Radiology
1995;196:439 43.
[2] Prince MR, Arnoldus C, Frisoli JK. Nephrotoxicity of
high-dose gadolinium compared with iodinated contrast. J Magn Reson Imaging 1996;1:162 6.
[3] Riederer SJ, Bernstein MA, Breen JF, et al. Threedimensional contrast-enhanced MR angiography with
real-time fluoroscopic triggering: design specifications
and technical reliability in 330 patient studies. Radiology 2000;215:584 93.
[4] Foo TKF, Saranathan M, Prince MR, et al. Automated
detection of bolus arrival and initiation of data acquisition in fast, three-dimensional, gadolinium-enhanced
MR angiography. Radiology 1997;203:275 80.
[5] Korosec FR, Frayne R, Grist TM, et al. Time-resolved
contrast-enhanced three-dimensional MR angiography.
[6] Earls JP, Rofsky NM, DeCorato DR, et al. Breath-hold
single-dose gadolinium-enhanced three-dimensional
MR aortography: usefulness of a timing examination
and MR power injector. Radiology 1996;201:705 10.
imaging with a volumetric interpolated breath-hold examination. Radiology 1999;212:876 84.
158
[8] Suto Y, Caner BE, Tamagawa Y, et al. Subtracted synthetic images in GD-DTPA enhanced MR. J Comput
Assist Tomogr 1989;13:925 8.
[9] Low RN, Martinez AG, Steinberg SM, et al. Potential
renal transplant donors: evaluation with gadoliniumenhanced MR angiography and MR urography. Radiology 1998;166:165 72.
[10] Motzer RJ, Bander NH, Nanus DM. Renal-cell carcinoma. N Engl J Med 1996;335:865 75.
[11] Semelka RC, Shoenlut JP, Magro CM, et al. Renal
cancer staging: comparison of contrast-enhanced CT
and gadolinium-enhanced fat-suppressed spin-echo
and gradient-echo MR imaging. J Magn Reson Imaging 1993;3:597 602.
[12] Wagner BJ, Wong-You-Cheong JJ, Davis Jr. CJ. From
the archives of the AFIP: adult renal hamartomas. Radiographics 1997;17:155 69.
[13] Helenon O, Serran S, Paraf F, et al. Unusual fat-containing tumors of the kidney: a diagnostic dilemma.
Radiographics 1997;17:129 44.
[14] Outwater EK, Bhatia M, Siegelman ES, et al. Lipid in
renal clear cell carcinoma: detection on opposedphase gradient-echo MR images. Radiology 1997;
205:103 7.
[15] Pickhardt PJ, Lonergan GJ, Davis CJ, et al. From the
archives of the AFIP. Infiltrative renal lesions: radiologic-pathologic correlation. Radiographics 2000;20:
215 43.
[16] Bosniak MA. Problems in the radiologic diagnosis of
renal parenchymal tumors. Urol Clin North Am
1993;20:217 30.
[17] Bosniak MA. The current radiological approach to renal cysts. Radiology 1986;158:1 10.
[18] Bosniak MA. Difficulties in classifying cystic lesion of
the kidney. Urol Radiol 1991;13:91 3.
[19] Bosniak MA. Diagnosis and management of patients
with complicated cystic lesions of the kidney. AJR Am
J Roentgenol 1997;169:819 21.
[20] Bosniak MA. The use of the Bosniak classification
system for renal cysts and cystic tumors. J Urol
1997;157:1852 3.
[21] Balci NC, Semelka RC, Patt RH, et al. Complex renal
cysts: Findings on MR imaging. AJR Am J Roentgenol
1999;172:1495 500.
[22] Semelka RC, Hricak H, Stevens SK, et al. Combined
gadolinium-enhanced and fat saturation MR imaging
of renal masses. Radiology 1991;178:803 9.
[23] Fishman EK, Deutch BM, Hartman DS, et al. Primary
adrenocortical carcinoma: CT evaluation with clinical
correlation. AJR Am J Roentgenol 1987;148:531 5.
[24] Katz RL, Shirkhoda A. Diagnostic approach to incidental adrenal nodules in the cancer patient. Cancer
1985;55:1995 2000.
[25] Reinig JW, Stutley JE, Leonhardt CM, et al. Differentiation of adrenal masses with MR imaging: comparison of techniques. Radiology 1994;192:41 6.
[26] Schwartz LH, Panicek DM, Koutcher JA, et al.
Adrenal masses in patients with malignancy: prospective comparison of echo-planar, fast spin echo,
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
and chemical shift MR imaging. Radiology 1995;

197:421 5.
Krestin GP, Friedmann G, Fischbach R, et al. Evaluation of adrenal masses in oncologic patients: dynamic
contrast-enhanced MR vs CT. J Comput Assist Tomogr
1991;15:104 10.
Chung J-J, Semelka RC, Martin DR. Adrenal adenomas: characteristic postgadolinium capillary blush on
dynamic MR imaging. J Magn Reson Imaging 2001;
13:242 8.
Outwater EK, Siegelman ES, Radecki PD, et al. Distinction between benign and malignant adrenal masses:
value of T1-weighted chemical shift MR imaging. AJR
Korobkin M, Lombardi TJ, Aisen AM, et al. Characterization of adrenal masses with chemical shift and
gadolinium-enhanced MR imaging. Radiology 1995;
197:411 8.
Schwartz LH, Macari M, Huvos AG, et al. Collision
tumors of the adrenal gland: demonstration and characterization at MR imaging. Radiology 1996;201:
757 60.
Outwater EK, Siegelman ES, Huang AB, et al. Adrenal masses: correlation between CT attenuation value
and chemical shift ratio at MR imaging with in-phase
and opposed-phase sequences. Radiology 1996;200:
749 52.
Caoili EM, Korobkin M, Francis IR, et al. Delayed
enhanced CT of lipid-poor adrenal adenomas. AJR
Korobkin M, Brodeur FJ, Francis IR, et al. Delayed
enhanced CT for differentiation of benign from malignant adrenal masses. Radiology 1996;200:737 42.
Pena CS, Boland GWL, Hahn PF, et al. Characterization of indeterminate (lipid poor) adrenal masses: use
of washout characteristics at contrast-enhanced CT.
Radiology 2000;217:798 802.
Boland GW, Goldberg MA, Lee MJ, et al. Indeterminate adrenal mass in patients with cancer: evaluation at
PET with 2-(F-18)-fluoro-2-deoxy-D-glucose. Radiology 1995;194:131 4.
Maurea S, Mainolfi C, Bazzicalupo L, et al. Imaging of
adrenal tumors using FDG PET: comparison of benign
and malignant lesions. AJR Am J Roentgenol
1999;173:25 9.
Dunnick NR, Heaston D, Halvorsen R, et al. CT appearance of adrenal cortical carcinoma. J Comput Assist Tomogr 1982;6:978 82.
Siegelman ES. MR imaging of the adrenal neoplasms.
Magon Reson Imaging Clin N Am 2000;4:769 86.
Kebebew E, Siperstein AE, Duh QY. Laparoscopic
adrenalectomy: the optimal surgical approach. J Laparoendosc Adv Surg Tech A 2001;11:409 13.
Bravo EL, Gifford RW. Current concepts: pheochromocytoma-diagnosis localization and management.
N Engl J Med 1984;311:1298 303.
Sutton MG, Sheps SG, Lie JT. Prevalence of clinically
unsuspected pheochromocytoma: review of a 50-year
autopsy series. Mayo Clin Proc 1981;56:354 60.

[43] Mayo-Smith WM, Boland GW, Noto RB, et al. Stateof-the-art adrenal imaging. Radiographics 2001;21:
995 1012.
[44] Beland SS, Vesely DL, Artnold WC, et al. Localization
of adrenal and extra-adrenal pheochromocytomas by
magnetic resonance imaging. South Med J 1989;82:
1410 3.
159
[45] Varghese JC, Hahn PF, Papanicolaou N, et al. MR

differentiation of pheochromocytoma from other adrenal lesions based on qualitative analysis of T2 relaxation times. Clin Radiol 1997;52:603 6.
[46] Kenney PJ, Wagner BJ, Rao P, et al. Myelolipoma:
CT and pathologic features. Radiology 1998;208:
87 95.
Radiol Clin N Am 41 (2003) 161 177
MR imaging of the bladder

Leo P. Lawler, MD, FRCR
The Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins Medical Institutions,
601 North Caroline Street, Room 3240F, Baltimore, MD 21287, USA
MR imaging of the bladder can now be performed

in a practical and time-efficient manner and has
entered the realm of routine body imaging. Particular
benefits of this modality include its tissue characterization, the particular response of the bladder wall to
fast dynamic contrast imaging, and the possibility of
pelvic floor functional studies without irradiation
[1,2]. Because of these benefits, MR imaging of the
bladder is encountered in many body-imaging protocols; however, the majority of patients currently
referred for dedicated bladder MR imaging are evaluated for primary bladder transitional cell carcinoma
or cystocele.
General techniques of bladder MR imaging

Patient preparation
The patient is imaged supine, feet first. A moderately distended bladder is preferred and is obtained by
asking the patient to void 2 to 3 hours before the
examination or, when present, clamping the Foley
catheter for a similar period [3]. An overdistended
bladder may decrease sensitivity for en-plaque wall
lesions and limit patient toleration. An underfilled
bladder may cause difficulty evaluating bladder thickening and cystocele. A smooth muscle relaxant, such
as 1 mg intramuscular glucagon, is administered to
decrease artifact from bowel movement. A waist
compression belt may be used to limit abdominal
wall motion artifact. Intravenous access for contrast is
required for bladder tumor studies. Double-phased
array surface pelvic coils allow for a small field of
E-mail address: llawler@jhmi.edu
view and improve signal-to-noise ratio [4 7]. Body

coils suffice for the upper abdominal retroperitoneal
and liver imaging.
Planes and sequences
A 1.5 T magnet provides the signal-to-noise and
contrast resolution for bladder wall and extravesical
detail. Stronger magnets facilitate the gradients and
resolution for fast, high-quality dynamic imaging and
breath-hold sequences. A three-plane gradient echo
defines coverage and assesses bladder distention
adequacy. A 28 cm to 32 cm axial field of view is
defined to include from the symphysis pubis to the
coccyx and, laterally, the lymphovascular structures
and ureters. Sagittal and coronal planes should cover
from the sacral promontory through the perineum.
The preferred slice thickness is 6 mm with a 2-mm
interslice gap. Initial scans include axial two-dimensional non fat-saturated conventional spin echo T1
(repetition time [TR] 400 600 milliseconds, echo
time [TE] minimum 20 milliseconds) and fat-saturated fast spin echo T2 (TR 4000 4500 milliseconds,
TE 80 100 milliseconds) sequences. If there is
concern for upper tract obstruction or retroperitoneal
adenopathy, axial single-shot fast spin echo (SSFSE)
studies with the body coil are added, with or without
coronal MR urography-type sequences.
Contrast-enhanced studies are performed with
gadolinium gadopentate administered by power
injector and two sequential, fat-saturated, gradient
echo T1 sequences (TR 4.7, TE minimum). Current
three-dimensional sequences provide contiguous
coverage with thinner slices over a large area while
maintaining high signal-to-noise and image contrast. Volume breath-hold examinations (eg, volumetic interpolated breath-hold examination) [8,9]
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 5 - 9
162
L.P. Lawler / Radiol Clin N Am 41 (2003) 161177
use interpolation to increase coverage, allowing for

160-mm slabs and 2.5-mm partitions in a 20 second
breath-hold. This provides speed and removes
interslice gaps as well as allowing for additional
planes to be reformatted after imaging.
Short tau inversion recovery sequences (STIR)
have utility for pelvic bone marrow and pelvic varices.
Gradient echo or postgadolinium angiogram studies
are used when pelvic vessel flow or compromise is a
question. Respiratory motion is not generally an issue
in the pelvis; however, in the presence of a large
pannus, an anterior fat saturation band or respiratory
gating may be of value [1]. Chemical shift artifact may
manifest in the frequency encode direction as a dark
line along one bladder wall and a high signal or absent
bladder wall on the contralateral side [10]. This is
seen with higher fields and is more prominent on
T2-weighted (T2W) sequences than on T1-weighted
(T1W) sequences. Fat saturation or changing the
frequency encode direction will help discriminate this
artifactual finding.
Breath-hold long TR sequences (half-Fourier
acquisition single-shot turbo spin echo [HASTE] or
SSFSE) or gradient echo sequences may be used to
generate virtual cystoscopic images through dedicated software [11]. Although this method is completely noninvasive, conventional cystoscopy is
considered minimally invasive and offers the possibility for immediate biopsy of suspicious findings.
sequences, the muscle has a slightly higher signal intensity than does urine and a slightly lower signal intensity than does perivesical fat. On T2W sequences,
it is contrasted between the high-signal urine and fat
[13]. Wall contrast improves on a T2W sequence,
with longer TE at the cost of signal to noise though.
The bright mucosa and lamina propria may be
obscured against the urine, however. Although not a
consistent finding, Maeda et al [14] have documented the three muscle layers on T2W sequences.
The superficial muscle layer has intermediate signal
intensity. The deep muscle layer has lower signal
intensity and is distinguishable using a T2W
sequence of detrusor hypertrophy [15]. The fetal
bladder can be imaged, in utero, using SSFSE
sequences in multiple planes.
Although the thin serosa is not perceptible as a
single layer, the outer aspect of the bladder wall is
defined against the high-signal perivesical fat on
T1W and T2W sequences. The ureterovesical junctions are best defined by the high-signal urine-filled
ureters on T2W or SSFSE sequences. Fascial planes
and lymph nodes are usually low signal on a T1W
sequence and are defined by their boundaries with
high-signal pelvic fat. The adjacent pelvic viscera
have internal signal characteristics that are well
described but, in general, are best depicted on non
fat-saturated T2. Normal vessels have signal voids on
spin echo sequences and the fatty marrow of pelvic
bones is defined by the surrounding low-signal cortex, which is best seen on a T1W sequence [1].
MR imaging of bladder anatomy

The bladder wall is comprised of four layers:
mucosa (uroepithelium), submucosa (lamina propria),
muscle layer (circular smooth muscle between inner
and outer longitudinal layers), and adventitia (connective tissue). A portion of the peritoneum serves as
bladder serosa by reinforcing the bladder dome. The
trigone has an extra layer of muscle and appears
slightly thicker than does the adjacent bladder wall.
The normal bladder wall, when moderately distended, is approximately 2 mm thick. Following the administration of gadolinium-diethylenetriaminepentaacetic acid, the normal bladder wall can show marked
enhancement of mucosa and lamina propria. The
lamina propria is composed of loose connective tissue
and abundant small vessels. Using the endorectal coil,
linear enhancement of the lamina propria is seen
following the administration of gadolinium [12].
Because of its long T1 relaxation time, the bladder
smooth muscle has intermediate signal intensity on
T1W sequences and low signal intensity on sequences with longer TR and TE. On moderate T1W
MR imaging of bladder neoplasms

Epidemiology of bladder tumors
The majority of bladder tumors are malignant and
the majority of urinary tract cancers are in the bladder,
making up about 2% of all cancers. They represent the
fourth most common cancer in men (6% to 8% of
men), with a 4 to 1 male to female distribution, a peak
in the sixth to seventh decade of life, and a rising
incidence in younger patients [16 19]. Approximately 90% are uroepithelial, and 90% of these are
transitional cell carcinoma. The remaining cell types
largely comprise squamous (5% 10% of bladder
tumors), adenocarcinoma (2% of bladder tumors),
and sarcoma cell types, as well as direct invasion
and metastases from other primary tumors. Many
tumors are of mixed cell type, although one type will
usually predominate. Nonuroepithelial neoplasms are
uncommon (less than 10%) and include rhabdomyosarcoma, non-Hodgkins lymphoma, plasmacytoma,
and pheochromocytoma. The single greatest risk factor for bladder cancer is cigarette smoking, causing a
sixfold increase in risk [20]. Other factors include
exposure to aniline dye, aromatic amine, diesel fume,
and phenacetin. A genetic predisposition has been
identified [21] and some chromosomal findings have
been linked to aggressiveness of the tumor [22].
Squamous tumors are linked to chronic cystitis and
schistosomiasis. Adenocarcinomas may invade from a
persistent urachus in the bladder wall supravesically,
or may occur in the trigone region [23,24].
Bladder cancer prognosis is predicated on depth
of invasion (T), nodal involvement (N), the presence
of metastatic foci (M), and histologic type and grade
(see Table 1) [20]. The TNM staging system has been
updated a number of times, the last time being in
1997 [20]. Deeply invading tumors are associated
with increased lymphatic spread and poorer prognosis
[25 28]. The various stages of bladder cancer affecting the bladder wall have been much discussed, but
have less impact on preoperative decision-making.
The main surgical demarcation is between organconfined and non organ-confined disease.
Table 1
The TNM classification for bladder cancer
Stage
Organ-confined
Ta
Tis
T1
Characteristic
disease
Noninvasive papillary carcinoma
Carcinoma in situ
Tumor invades subepithelial
connective tissue
T2 Tumor invades muscle)
T2a
Tumor invades superficial muscle
T2b
Tumor invades deep muscle
Non organ-confined disease
T3 Tumor invades perivesical tissue
T3a
Tumor invades perivesical
tissue microscopically
T3b
Tumor invades perivesical
tissue macroscopically (mass)
T4 Tumor invades adjacent viscera or body wall
T4a
Tumor invades prostate, uterus, or vagina
T4b
Tumor invades pelvic side wall or
abdominal wall
N0
No regional lymph node metastases
N1
Metastases to a single lymph node
V to 2 cm
N2
Metastases to a single lymph node > 2 cm
but V 5 cm, or multiple lymph nodes
No lymph nodes > 5 cm
N3
Metastasis in a lymph node > 5 cm
MO
No distant metastasis
M1
Distant metastasis
163
Role of MR imaging in bladder tumor evaluation

MR imaging is usually performed in patients who
already have a tumor diagnosed and localized. The
main function of MR imaging is to noninvasively
compliment the known shortcomings of clinical staging for tumors above stage T2a [23,29]. MR
imaging has been shown to provide additional prognostic information to clinical staging before radiation
therapy, where 30% of patients were upstaged and
20% were downstaged [30]. MR imaging is not used
for tumor detection [28,31], nor is it required for
differentiating histologic types of malignancy. The
standard of care for diagnosis of bladder cancer is
cystoscopy with deep muscle biopsy and a bimanual
examination to evaluate extravesical spread. The
accuracy of contrast-enhanced MR imaging in staging bladder cancer ranges from 50% to 96%
[3,18,32 36] and is 10% to 33% better than is CT
[2,8,13,26 28,33]. MR imaging has been found to be
particularly useful in differentiating stage T2 from
stage T3a and stage T3a from stage T3b.
MR imaging of the bladder is useful in other aspects of patient care of bladder cancer patients. MR
imaging is able to characterize extravesical growth
patterns, guide tissue sampling or neoadjuvant therapy, refine surgical approaches, and monitor response
and relapse after various therapeutic regimes. MR
imaging of the bladder is also able to assist in predicting and planning for the complications of advancing
disease. As an adjunct to conventional cystoscopy,
MR imaging of the bladder greatly facilitates multidisciplinary discussion of tumor management by providing a global imaging depiction of the tumor as it
relates to the pelvis and patient management.
MR imaging features of the primary bladder mass
On MR imaging, it is possible to see bladder
lesions as small as 1.5 cm [28,31,37,38]. MR imaging
distinguishes some characteristic features of bladder
tumors, but none are pathognomonic and benign and
malignant conditions cannot be reliably differentiated.
When transitional cell carcinomas occur, imaging
interpretation must be mindful that the entire uroepithelium is considered at risk for premalignant metaplasia and synchronous tumors. In addition, 33% of
patients have multifocal disease due to field cancerization or clonality [39,40]. Synchronous bladder
lesions occur in 2.3% of bladder cancer patients and
3.9% of patients with bladder lesions develop metachronous upper tract tumors; in contrast, 39% and
24% of patients with ureteral and renal tumors, respectively, will develop a bladder lesion (Fig. 1) [41].
164
Fig. 1. Axial, postcontrast, T1W image (FSPGR/70, TR

220/msec, TE 2/msec). Two synchronous transitional cell
carcinomas are at the left lateral wall (arrowheads) of the
bladder (B).
On T1W sequences, bladder carcinomas are of low

to intermediate signal intensity. They are isointense or
slightly hyperintense to muscle, hyperintense to urine,
and hypointense to perivesical fat (Fig. 2). On T2W
sequences, they are of intermediate to high signal
intensity. Bladder carcinomas are hyperintense to the
Fig. 2. Axial T1W image. The transitional cell carcinoma

(short arrow) is at the thickening of the left lateral bladder
wall (short arrow). Its intermediate signal is higher than that
of urine (U), although lower than that of perivesical fat ( F).
Note the fat plane (long arrow) between the seminal vesicles
(arrowheads) and the posterior bladder wall.
Fig. 3. Axial T2W image (FSE, TR 4000/msec, TE 77/

msec). The lobulated transitional cell carcinoma (T) signal is
higher than that of the muscle of the bladder wall (arrows),
although lower than that of the urine (B) and the perivesical
fat ( F).
bladder wall or late fibrosis, but are hypointense to

urine. Additionally, bladder carcinomas will have lower signal on fast spin echo sequences than on conventional spin echo sequences (Fig. 3). Enhancement also
occurs earlier than in edema or granulation tissue, but
postbiopsy or postirradiation changes and other causes
of inflammation can give false positive results unless
imaged over 3 months from the procedure [42].
Fig. 4. Axial T2W image (FSE, TR 4800/msec, TE

120/msec). A large papillary transitional cell carcinoma (T)
is seen arising from the posterolateral bladder (B). The lowsignal bladder wall is preserved (arrows) and the pedicle of
the tumor (arrowhead) is seen as a low-signal stalk.
Fig. 5. Axial, postcontrast, fat-saturated, T1W image. A

sessile transitional cell carcinoma (T ) involves most of the
anterior bladder (B) wall. Perivesical extension is suggested
by the indistinct margin with the anterior perivesical fat
(arrows), with enhancement noted within the saturated lowsignal perivesical fat.
Two thirds of malignant tumors have papillary

morphology and are found to be well differentiated
and invade superficially ( < stage T2) (Fig. 4). The
remaining one third present as solid, ulcerating, and
infiltrative masses that invade deeply or beyond the
bladder wall (>stage T2). These tend to have less welldifferentiated histology (Fig. 5) [16,43]. Polypoid
tumors have a pedicle of connective tissue, capillary
vessels, and inflammatory cells and edema; 75% of
these have an identifiable stalk that has a lower signal
than does the tumor on T2, less enhancement postgadolinium, and more enhancement on delayed
images, reflecting its fibrous nature with variable
vascular supply. Smooth muscle may be pulled up
Fig. 6. Axial, postcontrast, three-dimensional, gradient echo

image (FSPGR/70, TR 4.7/msec, TE 1.9/msec). An
increased linear enhancement (arrow) is the MR imaging
manifestation of this T1 tumor in the left posterior wall of
the bladder.
165
Fig. 7. Axial T2W image (FSE, TR 4317/msec, TE

53.3/msec). The intermediate signal tumor (arrowheads)
has breached the low-signal muscle (arrows) of the bladder
wall. The tumor also extends down from the dome of the
bladder (T).
into the stalk. Ninety percent of urachal carcinomas

arise juxtavesically, 6% arise from the middle segment, and 4% arise from the umbilical end [2]. Their
location and extent are often best appreciated on a
sagittal view that best illustrates the course of the
Fig. 8. Axial, postcontrast, fat-saturated, T1W image

(FSPGER, TR 270/msec, TE 4.2/msec). A tumor mass
(short arrow) enhances and extends beyond the normal
bladder wall (long arrow) into the perivesical fat ( F).
Normal, uninvolved, enhancing seminal vesicles (arrowwheads) are seen posterior to the bladder (B) and are
separated from it by the low-signal saturated fat.
166
Fig. 9. Axial, postcontrast, three-dimensional, fat-saturated,

gradient echo image (FSPGR/70, TR 4.8/msec, TE
1.9/msec). An asymmetric collection of small vessels
(arrows) was seen outside the bladder (B) wall adjacent the
tumor (T). No tumor invasion beyond the bladder was found
at surgery, however.
umbilicovesical fascia. The location is characteristic

and may suggest the diagnosis.
MR imaging features of direct tumor spread
MR imaging cannot reliably and consistently
distinguish between the early stages of bladder carcinoma (Tis, Ta, T1, and T2a), although this informa-
tion is usually discerned on biopsy. Distinction

between these early stages does not alter patient care.
Endorectal coils have been used to show disruption of
the submucosal connective tissue layer, but this
procedure is invasive and visualization is limited to
the posterior wall [12]. MR imaging may distinguish
the stage T2a tumor from the stage T2b tumor by
seeing the differential enhancement of tumor and
muscle or by recognizing a breach in the normal
low-signal muscle on a T2W sequence (Figs. 6, 7)
[16,17,27,28,31,38].
T3a spread can manifest as perivesical stranding,
wall irregularity, and retraction, but it may be hard to
differentiate from postbiopsy edema, perivesical
inflammation, or vasculature. After transurethral
resection, microperforations can lead to perivesical
changes and fat necrosis that can mimic tumor spread.
Extravesical extension (>stage T3a) is aided by differentiation from perivesical fat on postcontrast, highmatrix, T1W studies using subtraction techniques or
the application of fat saturation (Figs. 7, 8). Neovascularity may be seen beyond the bladder wall
adjacent to a tumor, but there is no data to suggest
this definitively implies a transmural component
(Fig. 9) and we have encountered cases in which it
did not correlate with deep invasion.
Muscle and viscera have shorter T2 relaxation
values than does tumor; therefore, T2 imaging will
maximize the contrast between them for distinguish-
Fig. 10. (A) Coronal T2W image (FSE, TR 3500/msec, TE 102/msec). Normal high-signal seminal vesicles (arrows) are seen
entering the prostate ( P). (B) Sagittal T2W image (FSE, TR 3500/msec, TE 102/msec). The normal high-signal fat plane (arrow) is
seen separating the seminal vesicles (S) from the low-signal posterior bladder wall (arrowheads). B, bladder; R, rectum. Loss of this
plane or the high signal within the seminal vesicle in the setting of posterior bladder tumor should arouse suspicion for extravesical
spread and seminal vesicle involvement.
167
Fig. 11. (A) Axial T2W image (FSE, TR 4167/msec, TE 77/msec). Low-signal tumor of the posterior bladder wall (arrows)
extends posteriorly around the seminal vesicles (arrowheads), obscures the high-signal fat plane normally present here, and
extends toward the rectum (R). B, bladder. (B) Axial, fat-saturated, T2W image (FSE, TR 3500/msec, TE 99/msec). The bilateral
hydronephrosis (arrowheads) with loss of renal cortex on the right (short arrow), although preserved on the right (long arrow), is
due to the posterior bladder tumor involvement of the ureteral orifices.
ing between T4a or T4b lesions. The T2W sequences

also better define any disruption of zonal anatomy
especially prostate and cervixand better delineate
any loss of fat planes demarcating the margins of
pelvic viscera and bone. Posterior tumor extension
into the seminal vesicles is suggested by focal areas
of decreased signal and by loss of the fat plane
separating the seminal vesicles from the bladder
anteriorly. Tubule wall enhancement after gadolinium
administration does not necessarily imply tumor
involvement [44]. Supplemental sagittal and coronal
planes are useful for imaging of prostate, perineum,
and lateral wall extension (Fig. 10).
Fecal discharge from the vagina or urethra or
pneumaturia should prompt a search for rectovesical
or rectovaginal fistula either due to tumor or therapy
such as irradiation. The former is best diagnosed by
the presence of low-signal bladder air on axial T2
sequences in a patient who has not been catheterized.
The latter is best discriminated on delayed sagittal,
T1W, fat-saturated images after administration of
intravenous contrast.
Hydroureter is circumstantial evidence of ureteral
orifice involvement. Increased uroepithelial enhancement within the ureter may be a nonspecific sign of
disease presence (Figs. 11, 12).
MR imaging and the role of contrast
Small tumor detection and staging accuracy is
improved with administration of intravenous contrast
[2,45]. Because of neovascularization, tumors take up

intravenous contrast with a more rapid rise to peak, a
higher peak, and a more rapid washout than do normal
tissue or granulation tissue. This results in an increase
in signal intensity and tumor conspicuity that is
Fig. 12. Axial, fat-saturated, postcontrast, subtraction, T1W

image (FSPGR/70, TR 255/msec, TE 4.2/msec). The patient
had multifocal transitional cell carcinoma of the bladder.
Enlargement and enhancement of the distal left ureter
(arrowhead) with periureteric stranding (arrow) was noted
and transitional cell carcinoma was found here at surgery,
necessitating further ureteric resection.
168
Fig. 13. (A) Coronal, postcontrast, fat-saturated, T1W image (FSPGR/70, TR 295/msec, TE 4.2/msec). A lobulated intermediate
signal transitional cell carcinoma (short arrow) hangs from the dome of the bladder (B). (B) Coronal, postcontrast, fat-saturated,
T1W image (FSPGR/70, TR 295/msec, TE 4.2/msec). On dynamic imaging, the mass (M) is noted to enhance briskly along with
thickening of the adjacent bladder dome (long arrow). The nonenhancing end of the tumor represents clot (arrowhead). The thin
normal bladder wall (short arrow) is noted. (C) Coronal, postcontrast, fat-saturated, subtraction, T1W image (FSPGR/70,
TR 295/msec, TE 4.2/msec). Subtraction increases the conspicuity of the enhancing tumor (long arrow) and confirms the lack of
enhancement of the clot (short arrow).
statistically significant [2,34,36,38,42,46 49]. Tumor

can be differentiated from normal wall as early as 5 to
15 seconds after contrast administration [3,34,50,51].
With delayed imaging, the contrast and signal in both
the tumor and the bladder wall decrease, and the high
signal of contrast in the bladder lumen may obscure a
small mucosal lesion. Subtraction images can increase
lesion conspicuity and may, on occasion, have greater
sensitivity for bladder lesions or organ involvement
(Figs. 12, 13).
Fast imaging is key to capturing the differential
enhancement times of mucosa, tumor, and inflam-
matory changes. Optimal tumor staging results from

using fast T1 sequences with short TR and TE and
multiple sections per breath hold following intravenous administration of 0.1 mmol/kg gadopentetate
dimeglumine [28,52]. Tanimoto et al [38] demonstrated that dynamic postgadolinium imaging improved
accuracy from 58% to 90% if small undetectable
tumors were excluded [38]. Volume acquisition with
three-dimensional imaging sequences may improve
staging by allowing customized postprocessing
planes, which may benefit lymph node detection
and local staging (Figs. 6, 9, 13) [38].
Fig. 14. Axial, postcontrast, fat-saturated, subtraction, T1W

image (FSPGR/70, TR 295/msec, TE 4.2/msec). A ringenhancing metastasis of primary bladder transitional cell
carcinoma was documented in the right pubic bone (arrow).
MR imaging and bone involvement

Bone metastasis is seen best on T1W images where
their low signal is similar to the primary tumor and is
contrasted against the fatty marrow background, or as
bright foci of high signal on fat-saturated T2W or
STIR images. Peripheral enhancement after gadolinium can be appreciated [2]. MR imaging documentation of direct extension toward or into the acetabulum
is of particular concern. Once this weight-bearing area
is involved, palliation is an extremely difficult orthopedic challenge. At the symphysis, differentiation of
radiation necrosis from tumor or infection is difficult.
Biopsy may be necessary, even though a very midline
and symmetrical pattern is unusual for tumor. If bone
involvement is detected, a bone scan is required to
assess the entire skeleton (Fig. 14) [8].
169
echo T1W studies [53]; high-resolution studies show

the nodes sharply circumscribed against the background fat and distinguishable from the signal voids
of the adjacent vessels. Nodes that are involved with
tumor may have normal signal characteristics and
size8 mm for a round node and 10 mm for an oval
node are considered the cut offs for malignancy. A
cluster of small nodes is also concerning. Morphology alone may give a positive predictive value of
94% for nodal involvement, a negative predictive
value of 89%, and an accuracy of 90% (Fig. 15)
[54]. Three-dimensional imaging may add some
information on lymph node shape but is unlikely to
add much clinically useful information, in contrast to
a close scrutiny of the routine sequences [54].
Contrast enhancement is nonspecific and insensitive with normal node morphology. Lymph nodespecific contrast agents containing ultrasmall iron
oxide particlessuch as ultrasmall superparamagnetic iron oxidehave been shown to accumulate
in benign nodes but not malignant ones [55]. Other
agents have been suggested to better demonstrate the
internal structure of nodes and their possible involvement with tumor [23,27,33,46,54,56].
Additional upper abdomen single-shot half-Fourier transform imaging with the body coil is performed
to assess upper abdomen retroperitoneal node status.
MR imaging-guided biopsy is feasible; however, in
its current form it is not practical. Once nodes are
MR imaging and lymph node disease

Bladder tumors that are less than stage T3 rarely
display lymph node involvement; however, higher
stages of bladder cancer demonstrate greater percentages of lymph node involvement. Lymph node
involvement generally encompasses the anterior and
lateral paravesical, presacral, hypogastric, obturator,
and external iliac nodes followed by the common
iliac and para-aortic lymph nodes.
MR imaging accuracy for nodal disease is 80% to
98%, which is similar to that for CT. Nodes are best
identified on non fat-saturated spin echo or fast spin

(FSPGR/70, TR 180, TE 4.2). A large metastatic lymph
node (short arrow) of primary bladder transitional cell
carcinoma is seen between the right external (arrowhead)
and internal (long arrow) vasculature.
170
MR imaging features of other bladder tumors

Bladder lymphoma
Bladder lymphoma is usually non-Hodgkins and
can affect the bladder as a primary process or
secondary involvement. It has been reported to present as thickening of the bladder with a signal intensity
that is intermediate on T1W and T2W sequences [41].
Benign mesenchymal tumors
Fig. 16. Sagittal, postcontrast, fat-saturated, T1W image

(FSPGR/70, TR 180/msec, TE 1.7/msec). This follow-up
study demonstrated contrast both within the neobladder
(arrow) and the vagina (arrowheads). Direct examination
confirmed a benign fistula, which was treated conservatively.
B, pubic bone.
localized by MR imaging, they are better biopsied by

conventional spiral CT or fluoro-CT.
Tumor treatment and follow-up
In general, superficial tumors ( < stage T2a) with
intact detrusor muscle are treated by transurethral
resection and adjunctive intravesical chemotherapy
such as BCG. Invasive tumors with limited perivesical spread require radical cystectomy and lymphadenectomy (>stage T2 and < stage T3b) with urinary
diversion or bladder reconstruction. Tumors between
these extremes can be considered for segmental
resection and bladder sparing if the entire tumor can
be confidently resected. Tumors that cannot be primarily resected may receive neoadjuvant therapy
with a view to downstaging for surgery or palliative
therapy. Five-year mortality for superficial tumors is
81% and for invasive tumors is 50% [57]. Following
radical cystectomy, long-term survival has been
reported to be 88% [47,57].
Fast dynamic MR imaging is the optimal modality
to demonstrate small, superficial lesions and the
vesical muscle layers. For this reason, it is the
preferred method used to follow-up lesions treated
conservatively by intravesical chemotherapy or transurethral resection [8,46]. Transitional cell carcinoma
can metastasize to the liver, where it can be detected
using dedicated liver sequences. Continued awareness of metachronous lesions in the remaining urinary
tract necessitates vigilance of the upper tracts on
follow-up examinations. MR imaging may also help
to evaluate postoperative complications (Fig. 16).
Mesenchymal tumors are rare, representing 0.23%

to 1% of bladder neoplasms (Fig. 17) [58]. All may
undergo malignant change. Leiomyomas are the most
common benign neoplasms of the bladder (0.04%
0.5% of bladder tumors) arising in the trigone near the
bladder outlet; they can grow into the lumen, in the
wall, or out through the bladder wall. These tumors
are intermediate on T1W sequences and intermediate
to low on T2W sequences. They are similar to tumors
in the uterus and show variable enhancement and
cystic change after contrast administration [59,60].
Neurofibromatas are either isolated or occur in
association with neurofibromatosis type 1. They are
usually situated in the ureter, trigone, or urethra. They
have intermediate to low signal characteristics on
T1W sequences, but have a higher signal than does
muscle. On T2W sequences they have an intermediate to high signal and are brighter than is fat with
marked contrast enhancement [59,61].
Ten percent of pheochromocytomas are extraadrenal and 1% of pheochromocytomas are in the
Fig. 17. Coronal, postcontrast, fat-saturated, T1W image

(FSPGER/70,TR 145/msec, TE 4.2/msec). A large mass
(arrow) in the dome of the bladder (B) was found to be
a plasmacytoma.
bladder, accounting for 0.06% of bladder tumors.

Pheochromocytomas occur at the trigone, ureteral
orifices, dome, and lateral walls (in that order), and
are situated in the submucosa or within the wall. These
lesions have been described as their adrenal counterparts: isointense on T1W sequences and homogenously hyperintense on T2W imaging. The tumors
may be obscured by the high-signal urine on T2
sequences, but they show marked enhancement after
gadolinium [59,62].
The bladder and prostate of children is the most
common site for rhabdomyosarcoma, an aggressive
tumor of the genitourinary tract. This tumor has been
reported as isointense on T1W sequences and hyperintense to urine and wall on proton density; contrast
enhancement also is seen [63]. Multiplanar imaging is
of value to characterize the often-widespread extension beyond the confines of the pelvis and perineum.
Inflammatory pseudosarcoma or pseudotumor is
an uncommon spindle cell neoplasm of a polypoid
nature, with signal characteristics reported of low
T1W signal and inhomogenous high T2W signal, as
well as inhomogenous enhancement [58,59]. This
lesion runs a benign course, but has a locally aggressive appearance that can mimic a tumor such as a rhabdomyosarcoma. It is of low signal intensity on T1W
and T2W sequences, is highly vascular, may be
associated with clot, and demonstrates marked contrast enhancement.
Inflammatory pseudosarcoma or pseudotumor is
an uncommon spindle cell neoplasm of a polypoid
nature. Its signal characteristics are reportedly low on
T1W sequences and inhomogenously high on T2W
Fig. 18. Axial, single-shot fast spin echo image (SSFSE, TR

47468/msec, TE 98.5/msec). A wide-mouthed anterior
bladder diverticulum (arrowhead ) is seen to extend through
diastasis of the recti muscles (arrows).
171

(FSPGR/70, TR 235/msec, TE 4.2/msec) shows a narrownecked (arrow) posterior diverticulum with increased
enhancement of the mucosa (arrowheads), which was found
to be TCC.
sequences [58,59]. This lesion runs a benign course,

but has a locally aggressive appearance that can
mimic a tumor such as a rhabdomyosarcoma. It is
highly vascular, may be associated with clot, and
demonstrates marked contrast enhancement.
MR imaging of other bladder conditions

Bladder diverticula represent a weakness in the
bladder wall through which mucosa penetrates. Seven
percent of bladder tumors occur within diverticula,

(FSPGR/70, TR 200/msec, TE 1.7/msec). A narrow neck
(arrow) leads into a large posterior diverticulum (T) from
the posterior bladder (B). There is increased mucosal
enhancement (arrowheads), which was found to be transitional cell carcinoma.
172
Fig. 21. Axial, fat-saturated, T2W image (FSE, TR

5966/msec, TE 96/msec). A trabeculated bladder (B) gives
rise to a left anterolateral diverticulum (T), which contains a
mass of transitional cell carcinoma (T ).
and are thought to be related to, stasis and chronic

inflammation [20]. Because these diverticula do not
contain all three muscle layers, they are thought to be
beyond stage T2b at diagnosis. MR imaging has a lot
to offer in the diagnosis and management of these
tumors. In the absence of calcification, any soft tissue
within a diverticulum should be treated as a tumor
until proven otherwise. MR imaging interpretation of
the size and site of the diverticulum neck as well as
any extension of tumor into perivesical tissues may
be valuable to the urologist because it can be difficult
Fig. 22. Axial, fat-saturated, T1W image (FSPGR/70, TR

270/msec, TE 4.2/msec). Circumferential increased superficial enhancement of the bladder mucosa (arrowheads) is
due to BCG-induced cystitis.
Fig. 23. Axial, fat-saturated, T2W image (FSE TR

4402/msec, TE 65/msec). High-signal urine extends through
a dissectionlike disruption and flap (black arrow) of the left
lateral bladder wall (white arrows), which were thought to
be related to the patients prior exposure to schistosomiasis.
to find the orifice or negotiate through it at cystoscopy (Figs. 18 21)

Vesical congestion and inflammation are seen as
high-signal areas, possibly with associated wall thickening, compared with the low-signal T2 of the
remaining bladder. Both tumor and inflammation
lengthen T1 and T2 values of bladder. Enhancement
is variable. If there is a hemorrhagic component it
Fig. 24. Sagittal T2W image (FSE, TR 4616/msec, TE

96/msec). This midline image demonstrates the absence of
the high-signal urine-filled bladder (arrowhead) behind the
symphysis (short arrow) in this infant with bladder
extrophy. There is a protuberance of soft tissue (long arrow)
through the anterior abdominal wall defect.
173
urinary habit disturbance causing significant distress

and lifestyle disruption. Pelvic floor prolapse is a
complex problem involving the interrelationship of
the bladder, bowel, and vaginal vault, their supporting
musculature, surrounding fascia layers, and innervation. Symptoms and the physical examination may be
limited in terms of defining what is actually prolapsing
and its predisposing causes. When there is also uterine
prolapse and organs fill the introitus, direct examination becomes more difficult.
Normal anatomy and cystocele
Fig. 25. Axial, single-shot fast spin echo image (SSFSE,
TR 60250/msec, TE 97/msec). A normal, in-utero, urinefilled bladder is documented (B) between the lower limbs
(white arrows). Amniotic fluid (black arrow) and placenta
( P) are seen.
may be of high signal intensity on T1 sequences [2].

Infection, radiation, and postbiopsy changes are
among the most common etiologies (Figs. 22, 23).
Endometriosis affects 10% of menstruating women. The urinary tract is the site of involvement in 1%
of cases. Of these, 84% affect the bladder, usually at
the posterior wall and dome. Implantation is thought
to occur at surgery, as a result of extensive pelvic
endometriosis and possibly due to direct extension of
adenomyosis. Symptoms involve pain, dysuria, and
hematuria. Implants are hemorrhagic and cystic and
their signal characteristics may vary depending on the
amount and timing of hemorrhage. High-signal T1W
and variably high or low T2W signal characteristics
have been seen.
The absence of the bladder and diastasis of the
symphysis pubis are the features observed on T2W
sequences of bladder extrophy (Fig. 24). This also
can be suggested on in-utero imaging (Fig. 25).
Anterior placenta percreta can be observed best
on sagittal T2W sequences, and any disruption on
the normal uterovesical plane can be documented
together with extension of the placenta through the
uterine wall.
Cystocele
The pelvic floor serves to support the pelvic organs
and aid their normal functions of controlled continence, urination, and defecation. This support may be
compromised through physical insult such as childbirth and the loss of the estrogenic support of the tissue
integrity. Pelvic floor weakness and organ prolapse is
common in women and can lead to pain, pressure, and
The bladder is normally supported by the levator

muscles, pubococcygeus, puborectalis, and endopelvic fascia. The anterior levator muscle has a slinglike
configuration enclosing and supporting the urogenital
hiatus. Vesicopelvic, urethropelvic, and pubourethral
ligaments and fascia give anterolateral support of the
bladder neck and urethra through attachment to the
pubic bone and arcus tendineus fasciae [64]. The pubocervical fascia forms a septum to support the
posterior bladder wall against the anterior vagina.
Bladder prolapse or cystocele is due to failure of this
fascial support with or without associated perineal
descent. There is posteroinferior displacement of the
bladder, causing anterior bulging of the vaginal vault.
The severity of prolapse is graded relative to the
hymen. Symptoms include pressure, discomfort, and
distortion of the urethra. Prolapse can lead to urinary
outflow problems and secondary infection [65].
Cystocele and the role of MR imaging
The goal of MR imaging is to document normal
anatomy and correlate clinical findings with defined
abnormal anatomy and function, to aid precision in
surgical planning and monitor response to therapeutic
intervention. Pelvic MR imaging tissue contrast is
well suited to documenting the anatomy of the urinefilled bladder and its relationship to the other organs
and supporting structures of the pelvic floor, both at
rest and during straining. Pelvic irradiation should be
avoided in younger women. Pathological MR
imaging findings have found good correlation with
pelvic organ prolapse and urinary dysfunction, and
are complimentary to dynamic cystocoloproctography and clinical examination [66 68]
Technique
Patients are usually imaged supine with hips and
knees flexed to simulate the sitting position [65,67,
69,70]. There are some advocates of imaging in up-
174
right scanners to better simulate the effects of gravity

on pelvic floor dysfunction, and indeed the findings
are thought to be quite different between supine and
upright scans [66,71,72]. A pelvic phased-array coil
and SSFSE sequence is used (infinite TR; TE 60 milliseconds, HASTE or SSFSE), with 5-mm section
thickness, and 1-mm gap. Axial, sagittal, and coronal
planes are utilized, although sagittal imaging is best
for demonstrating the interrelationship of the pelvic
floor structures. Images are acquired during rest and
strain. Strain imaging is acquired with either the
Valsalva maneuver or actual evacuation of contrast
material. Dynamic image review for interpretation of
the pelvic floor with cine-loop display has been

effective in demonstrating pelvic organ prolapse
[70] and allowing for judgment of the interrelationship of moving structures with respect to fixed anatomical landmarks.
The basic examination requires no contrast and
will often suffice for cystocele evaluation alone.
Options for contrast material include bladder, small
bowel, rectal, and vaginal contrast [69,70,73]. Oral
and rectal contrast choices include positive high T1W
and T2W signal ferric ammonium citrate or high
T1W gadopentate dimeglumine [56,69,70,73,]. US
gel and potatoes mixed with gadopentate dimeglu-
Fig. 26. (A) Sagittal T2W image (SSFSE, TR 2195/msec, TE 69/msec). On this resting image, urine is in the bladder (long
arrow) and there is high-signal contrast in the vagina (arrowhead) and rectum (short arrow). (B) Sagittal single-shot fast spin
echo image (SSFSE, TR 30711/msec,TE 69/msec) during straining. The bladder is seen to prolapse posteroinferiorly (long
arrow) behind the pubic bone ( P). An anterior rectocele (short arrow) and contrast from the vagina (arrowhead) are also noted.
(C) Axial single-shot fast spin echo image (SSFSE, TR 23311/msec,TE 68/msec). This inferior view demonstrates the cystocele
during straining as it protrudes through the introitus (arrow).
175
(ie, Kegel exercises). Surgical approaches involve

bladder suspension and reconstruction of the pubocervical fascia and inferior supporting tissues. Failure
of therapy is not uncommon and may be due to
failure to fully appreciate the anatomic derangements
being treated [71]. If problems are due to failure of
muscle trophic therapies, exercise may help; however, when fascial defects are documented, surgical
repair is often indicated [71].
Summary
Fig. 27. Sagittal T2W image (FSE, TR 2386/msec, TE

69/msec). A large cystocele (C) is seen to prolapse below the
pubic bone ( P). The weak pelvic floor also contributes to a
rectal prolapse (R).
mine have been used as rectal contrast [69,72].

Negative bowel agents include supermagnetic iron
particles, diamagnetic barium sulfate, and air. Markers may be placed at the anus, and urethral catheters
can be used for its localization.
MR imaging of the bladder can now be routinely

and easily performed with consistent quality. The
latest techniques provide high-resolution images of
the soft tissue contrast and the ability to perform
dynamic contrast imaging and functional pelvic floor
studies. MR imaging fits in as a complimentary tool
to cystoscopy and to conventional pelvic floor
dynamic studies and can provide a unique imaging
perspective of the bladder. It is important, however, to
remain cognizant of the limitations of its use in tumor
detection and discrimination.
References
MR imaging features of cystocele
The normal bladder neck at MR imaging should be
close to the symphysis, with an H-shaped vaginal
lumen and horizontal urethropelvic ligaments [74].
On MR imaging, the posterior bladder wall forms a
smooth bulge that extends posteroinferiorly. Its apex is
directed toward the introitus, whereas the relatively
fixed anterior bladder remains undisturbed. The exact
configuration of the bulge will depend somewhat on
the site and character of any pubocervical fascia defect.
Associated descent of the cervix and remaining pelvic
floor may alter the configuration of the cystocele. The
imaging definition of cystocele has been variably
described as bladder below the symphysis, below the
pubococcygeal line, and 1 cm below the pubococcygeal line. Loss of muscle bulk and increased T1W
signal may help to document predisposing fatty
atrophy, and follow-up studies may help to monitor
any trophic effects of estrogenic therapy (Figs. 26, 27).
Cystocele treatment
Conservative therapies are aimed at restoring the
firmness of the pelvic floor and include hormonal
therapy, pessaries, and pelvic floor muscle exercises
[1] Banson ML. Normal MR anatomy and techniques for

imaging of the male pelvis. Magn Reson Imaging Clin
N Am 1996;4(3):481 96.
[2] Teeger S, Sica GT. MR imaging of bladder diseases.
Magn Reson Imaging Clin N Am 1996;4(3):565 81.
[3] Barentsz JO, Ruijs SH, Strijk SP. The role of MR
imaging in carcinoma of the urinary bladder. AJR
Am J Roentgenol 1993;160(5):937 47.
[4] Barentsz JO, Lemmens JA, Ruijs SH, et al. Carcinoma
of the urinary bladder: MR imaging with a double surface coil. AJR Am J Roentgenol 1988;151(1):107 12.
[5] Hayes CE, Dietz MJ, King BF, Ehman RL. Pelvic
imaging with phased-array coils: quantitative assessment of signal-to-noise ratio improvement. J Magn
[6] Smith RC, Reinhold C, Lange RC, McCauley TR, Kier
R, McCarthy S. Fast spin-echo MR imaging of the
female pelvis. Part I. Use of a whole-volume coil. Radiology 1992;184(3):665 9.
[7] Smith RC, Reinhold C, McCauley TR, et al. Multicoil
high-resolution fast spin-echo MR imaging of the female pelvis. Radiology 1992;184(3):671 5.
[8] Barentsz JO, Witjes JA, Ruijs JH. What is new in bladder cancer imaging. Urol Clin North Am 1997;24(3):
583 602.
imaging with a volumetric interpolated breath-hold examination. Radiology 1999;212(3):876 84.
176
[10] Lee JK, Rholl KS. MRI of the bladder and prostate.
AJR Am J Roentgenol 1986;147(4):732 6.
[11] Bernhardt TM, Rapp-Bernhardt U. Virtual cystoscopy
of the bladder based on CT and MRI data. Abdom
Imaging 2001;26(3):325 32.
[12] Hayashi N, Tochigi H, Shiraishi T, Takeda K, Kawamura J. A new staging criterion for bladder carcinoma
using gadolinium-enhanced magnetic resonance imaging with an endorectal surface coil: a comparison with
ultrasonography. BJU Int 2000;85(1):32 6.
[13] Fisher MR, Hricak H, Crooks LE. Urinary bladder MR
imaging. Part I. Normal and benign conditions. Radiology 1985;157(2):467 70.
[14] Maeda H, Kinukawa T, Hattori R, Toyooka N, Furukawa T, Kuhara H. Detection of muscle layer invasion
with submillimeter pixel MR images: staging of bladder
carcinoma. Magn Reson Imaging 1995;13(1):9 19.
[15] Narumi Y, Kadota T, Inoue E, et al. Bladder wall morphology: in vitro MR imaging histopathologic correlation. Radiology 1993;187(1):151 5.
[16] Husband JE. Bladder cancer. In: Husband JE, Reznek
RH, editors. Imaging in oncology. 1st ed. Oxford
(UK): Virginia ISIS Medical Media; 1998. p. 215 39.
[17] Husband JE. Staging bladder cancer. Clin Radiol 1992;
46(3):153 9.
[18] Husband JE, Olliff JF, Williams MP, Heron CW, Cherryman GR. Bladder cancer: staging with CT and MR
imaging. Radiology 1989;173(2):435 40.
[19] Jewett MA, Pereira G, Nijmeh P, et al. Increasing incidence, but stable mortality, of bladder cancer in Ontario. Analysis of a population sample. Urology 1991;
37(Suppl):4 7.
[20] MacVicar AD. Bladder cancer staging. BJU Int 2000;
86(Suppl 1):111 22.
[21] Lynch HT, Kimberling WJ, Lynch JF, Brennan K.
Familial bladder cancer in an oncology clinic. Cancer
Genet Cytogenet 1987;27(1):161 5.
[22] Presti Jr JC, Reuter VE, Galan T, Fair WR, CordonCardo C. Molecular genetic alterations in superficial
and locally advanced human bladder cancer. Cancer
Res 1991;51(19):5405 9.
[23] Barentsz JO, Jager GJ, Witjes JA. MR imaging of the
urinary bladder. Magn Reson Imaging Clin N Am
2000;8(4):853 67.
[24] Rafal RB, Markisz JA. Urachal carcinoma: the role of
magnetic resonance imaging. Urol Radiol 1991;12(4):
184 7.
[25] Buy JN, Moss AA, Guinet C, et al. MR staging of
bladder carcinoma: correlation with pathologic findings. Radiology 1988;169(3):695 700.
[26] Fisher MR, Hricak H, Tanagho EA. Urinary bladder
MR imaging. Part II. Neoplasm. Radiology 1985;
157(2):471 7.
[27] Rholl KS, Lee JK, Heiken JP, Ling D, Glazer HS.
Primary bladder carcinoma: evaluation with MR imaging. Radiology 1987;163(1):117 21.
[28] Tachibana M, Baba S, Deguchi N, et al. Efficacy of
gadolinium-diethylenetriaminepentaacetic acid-enhanced magnetic resonance imaging for differentiation
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
between superficial and muscle-invasive tumor of the

bladder: a comparative study with computerized tomography and transurethral ultrasonography. J Urol
1991;145(6):1169 73.
Siegelman ES, Schnall MD. Contrast-enhanced MR
imaging of the bladder and prostate. Magn Reson
Imaging Clin N Am 1996;4(1):153 69.
Robinson P, Collins CD, Ryder WD, et al. Relationship
of MRI and clinical staging to outcome in invasive
bladder cancer treated by radiotherapy. Clin Radiol
2000;55(4):301 6.
Tanimoto A, Yuasa Y, Endo M, et al. [Magnetic resonance imaging of bladder tumors: superiority of serial
fast SE assisted by Gd-DTPA in tumor staging].
Nippon Igaku Hoshasen Gakkai Zasshi 1989;49(12):
1552 66.
Amendola MA, Glazer GM, Grossman HB, Aisen
AM, Francis IR. Staging of bladder carcinoma:
MRI-CT-surgical correlation. AJR Am J Roentgenol
1986;146(6):1179 83.
Bryan PJ, Butler HE, LiPuma JP, Resnick MI, Kursh
ED. CT and MR imaging in staging bladder neoplasms. J Comput Assist Tomogr 1987;11(1):96 101.
Kim B, Semelka RC, Ascher SM, Chalpin DB, Carroll
PR, Hricak H. Bladder tumor staging: comparison of
contrast-enhanced CT, T1- and T2- weighted MR
imaging, dynamic gadolinium-enhanced imaging, and
late gadolinium-enhanced imaging. Radiology 1994;
193(1):239 45.
Narumi Y, Kadota T, Inoue E, et al. Bladder tumors:
staging with gadolinium-enhanced oblique MR imaging. Radiology 1993;187(1):145 50.
Neuerburg JM, Bohndorf K, Sohn M, Teufl F,
Guenther RW, Daus HJ. Urinary bladder neoplasms:
evaluation with contrast-enhanced MR imaging. Radiology 1989;172(3):739 43.
Narumi Y, Kumatani T, Sawai Y, et al. The bladder and
bladder tumors: imaging with three-dimensional display of helical CT data. AJR Am J Roentgenol 1996;
167(5):1134 5.
Tanimoto A, Yuasa Y, Imai Y, et al. Bladder tumor
staging: comparison of conventional and gadoliniumenhanced dynamic MR imaging and CT. Radiology
1992;185(3):741 7.
Aben KK, Witjes JA, van Dijck JA, Schalken JA,
Verbeek AL, Kiemeney LA. Lower incidence of urothelial cell carcinoma due to the concept of a clonal
origin. Eur J Cancer 2000;36(18):2385 9.
Murphy WM, Nagy GK, Rao MK, et al. Normal
urothelium in patients with bladder cancer: a preliminary report from the National Bladder Cancer Collaborative Group A. Cancer 1979;44(3):1050 8.
Yousem DM, Gatewood OM, Goldman SM, Marshall
FF. Synchronous and metachronous transitional cell
carcinoma of the urinary tract: prevalence, incidence,
and radiographic detection. Radiology 1988;167(3):
613 8.
Barentsz JO, Jager GJ, van Vierzen PB, et al. Staging
urinary bladder cancer after transurethral biopsy: value
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
of fast dynamic contrast-enhanced MR imaging. Radiology 1996;201(1):185 93.

Kiemeney LA, Witjes JA, Verbeek AL, Heijbroek RP,
Debruyne FM. The clinical epidemiology of superficial
bladder cancer. Dutch South East Cooperative Urological Group. Br J Cancer 1993;67(4):806 12.
Secaf E, Nuruddin RN, Hricak H, McClure RD, Demas B. MR imaging of the seminal vesicles. AJR Am J
Roentgenol 1991;156(5):989 94.
Sparenberg A, Hamm B, Hammerer P, Samberger V,
Wolf KJ. [The diagnosis of bladder carcinomas by
NMR tomography: an improvement with Gd-DTPA?].
Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr
1991;155(2):117 22.
Barentsz JO, Engelbrecht M, Jager GJ, et al. Fast dynamic gadolinium-enhanced MR imaging of urinary
bladder and prostate cancer. J Magn Reson Imaging
1999;10(3):295 304.
Cheng L, Weaver AL, Leibovich BC, et al. Predicting
the survival of bladder carcinoma patients treated with
radical cystectomy. Cancer 2000;88(10):2326 32.
Neuerburg JM, Bohndorf K, Sohn M, Teufl F, Gunther
RW. Staging of urinary bladder neoplasms with MR
imaging: is Gd-DTPA helpful? J Comput Assist Tomogr 1991;15(5):780 6.
Sohn M, Neuerburg J, Teufl F, Bohndorf K. Gadolinium-enhanced magnetic resonance imaging in the
staging of urinary bladder neoplasms. Urol Int
1990;45(3):142 7.
Barentsz JO, Ruijs SH, van Erning LJ. Magnetic resonance imaging of urinary bladder cancer: an overview
and new developments. Magn Reson Q 1993;9(4):
235 58.
Cheng D, Tempany CM. MR imaging of the prostate
and bladder. Semin Ultrasound CT MR 1998;19(1):
67 89.
Barentsz JO, Jager GJ, Witjes JA, Ruijs JH. Primary
staging of urinary bladder carcinoma: the role of MRI
and a comparison with CT. Eur Radiol 1996;6(2):
129 33.
Heiken JP, Lee JK. MR imaging of the pelvis. Radiology 1988;166(1, Pt 1):11 6.
Jager GJ, Barentsz JO, Oosterhof GO, Witjes JA,
Ruijs SJ. Pelvic adenopathy in prostatic and urinary
bladder carcinoma: MR imaging with a three-dimensional TI-weighted magnetization-prepared- rapid gradient-echo sequence. AJR Am J Roentgenol 1996;
167(6):1503 7.
Weissleder R, Elizondo G, Wittenberg J, Lee AS, Josephson L, Brady TJ. Ultrasmall superparamagnetic
iron oxide: an intravenous contrast agent for assessing
lymph nodes with MR imaging. Radiology 1990;
175(2):494 8.
Scheidler J, Reiser MF. MRI of the female and male
pelvis: current and future applications of contrast enhancement. Eur J Radiol 2000;34(3):220 8.
Rozanski TA, Grossman HB. Recent developments in
the pathophysiology of bladder cancer. AJR Am J
Roentgenol 1994;163(4):789 92.
177
[58] Fujiwara T, Sugimura K, Imaoka I, Igawa M. Inflammatory pseudotumor of the bladder: MR findings.
J Comput Assist Tomogr 1999;23(4):558 61.
[59] Chen M, Lipson SA, Hricak H. MR imaging evaluation of benign mesenchymal tumors of the urinary
bladder. AJR Am J Roentgenol 1997;168(2):399 403.
[60] Maya MM, Slywotzky C. Urinary bladder leiomyoma:
magnetic resonance imaging findings. Urol Radiol
1992;14(3):197 9.
[61] Shonnard KM, Jelinek JS, Benedikt RA, Kransdorf
MJ. CT and MR of neurofibromatosis of the bladder.
[62] Warshawsky R, Bow SN, Waldbaum RS, Cintron J.
Bladder pheochromocytoma with MR correlation.
[63] Fletcher BD, Kaste SC. Magnetic resonance imaging
for diagnosis and follow-up of genitourinary, pelvic,
and perineal rhabdomyosarcoma. Urol Radiol 1992;
14(4):263 72.
[64] Stoker J, Halligan S, Bartram CI. Pelvic floor imaging.
Radiology 2001;218(3):621 41.
[65] Pannu HK, Kaufman HS, Cundiff GW, Genadry R,
Bluemke DA, Fishman EK. Dynamic MR imaging of
pelvic organ prolapse: spectrum of abnormalities. Radiographics 2000;20(6):1567 82.
[66] Fielding JR, Griffiths DJ, Versi E, Mulkern RV, Lee
ML, Jolesz FA. MR imaging of pelvic floor continence
mechanisms in the supine and sitting positions. AJR
Am J Roentgenol 1998;171(6):1607 10.
[67] Huddleston HT, Dunnihoo DR, Huddleston 3rd PM,
Meyers Sr PC. Magnetic resonance imaging of defects
in DeLanceys vaginal support levels I, II, and III. Am
J Obstet Gynecol 1995;172(6):1778 84.
[68] Kaufman HS, Buller JL, Thompson JR, et al. Dynamic
pelvic magnetic resonance imaging and cystocolpoproctography alter surgical management of pelvic floor
disorders. Dis Colon Rectum 2001;44(11):1575 84.
[69] Lienemann A, Anthuber C, Baron A, Kohz P, Reiser
M. Dynamic MR colpocystorectography assessing pelvic-floor descent. Eur Radiol 1997;7(8):1309 17.
[70] Yang A, Mostwin JL, Rosenshein NB, Zerhouni EA.
Pelvic floor descent in women: dynamic evaluation
with fast MR imaging and cinematic display. Radiology 1991;179(1):25 33.
[71] Hoyte L, Versi E. Magnetic resonance imaging and the
pelvic floor. In: Pemberton J, Swash M, Henry MM,
editors. The pelvic floor. Its function and disorders.
London: W. B. Saunders; 2002. p. 466 77.
[72] Schoenenberger AW, Debatin JF, Guldenschuh I, Hany
TF, Steiner P, Krestin GP. Dynamic MR defecography
with a superconducting, open-configuration MR system. Radiology 1998;206(3):641 6.
[73] Healy JC, Halligan S, Reznek RH, et al. Magnetic resonance imaging of the pelvic floor in patients with
obstructed defaecation. Br J Surg 1997;84(11):1555 8.
[74] Klutke C, Golomb J, Barbaric Z, Raz S. The anatomy
of stress incontinence: magnetic resonance imaging of
the female bladder neck and urethra. J Urol 1990;
143(3):563 6.
Radiol Clin N Am 41 (2003) 179 192
MR imaging of the female pelvis

Julia R. Fielding, MD
Department of Radiology, University of North Carolina Chapel Hill, Campus Box 7510, University of North Carolina Hospital,
101 Manning Drive, Chapel Hill, NC 27599, USA
MR imaging is becoming an increasingly important tool in the diagnosis of benign and malignant
disease of the female pelvis. Although ultrasound
(US) and hysterosalpingography (HSG) remain the
primary forms of imaging, MR imaging is now
routinely used in the diagnostic work-up of infertility,
including mullerian anomalies, and chronic pelvic
pain. Both CT and MR imaging can be used to
determine the origin of and to characterize a pelvic
mass. Staging of gynecologic malignancies can also
be performed with both modalities. In obstetrics, MR
imaging is used to assess maternal complications of
pregnancy and to identify or confirm fetal anomalies.
MR imaging technique
Before imaging, the patient should fast for 6 hours
to diminish bowel peristalsis. Alternatively, 0.5 to
1 mg of glucagon can be administered intramuscularly
at the beginning of the examination. It is also advisable to have the patient void before the examination to
limit deformation of adjacent organs by an enlarged
bladder. Although imaging using the body coil is
certainly adequate for most diagnoses, the highest
resolution images are obtained using a multicoil array.
All major vendors have such an array in which several
surface coils are enclosed within a band that encircles
the pelvis. Information from the surface coils is
summed to form the final images. Saturation bands
placed along the anterior and posterior body wall fat
are useful to diminish near field artifact.
Coronal, rapid T2-weighted scout images encompassing the entirety of the abdomen and pelvis serve
E-mail address: Julia_fielding@med.unc.edu
two purposes: (1) identification of the central pelvis

to optimize centering of subsequent small field-ofview images and (2) identification of the kidneys
to exclude anomalies and hydronephrosis. Smaller
field-of-view sagittal, axial, and sometimes coronal
T2-weighted images are then obtained because they
contain the most information about the pelvic organs.
A pulse sequence with simultaneous encoding of
multiple 180 pulses, such as fast or turbo spin echo,
yields the highest resolution images. Twenty-centimeter field-of-view, 128 phase-encoding steps, echo
train length of 8, and slices 3 to 5 mm in thickness
with minimal gap are good general sequence parameters. Ultra-fast T2-weighted images, such as half
Fourier single-shot turbo spin echo or single shot fast
spin echo (SSFSE), may be adequate, but require
larger fields-of view and hence yield decreased spatial resolution.
After all important T2 imaging is obtained, the
need for further imaging is variable. In the case of
benign disease of the uterus, no further imaging is
necessary. When characterizing a pelvic mass or
staging a pelvic malignancy, dynamic, axial, or sagittal images of the pelvis should be obtained at 20 seconds, 40 seconds, and 3 to 5 minutes following the
administration of intravenous contrast medium. Fat
saturation, although not essential, is useful for highlighting areas of enhancement.
Normal anatomy
Because of the unique contrast differences among
the viscera, T2-weighted MR images are often superior
to CT and US in assessing the female pelvis. The uterus
has three distinct zones: (1) the high T2 signal endometrium, (2) the low T2 signal junctional zone, and (3)
PII: S 0 0 3 3 - 8 3 8 9 ( 0 2 ) 0 0 0 6 3 - 5
180
J.R. Fielding / Radiol Clin N Am 41 (2003) 179192
Fig. 1. Sagittal (A) and axial (B,C) T2-weighted images of the normal female pelvis. On the sagittal image (A), the three zones of the
uterus are demarcated: (1) the myometrium (black arrows); (2) the junctional zone (white arrowheads); and (3) the bright
endometrium (e). Incidental note is made of a bladder fold. (B) The normal convex fundal contour (short arrows) and the normal
appearance of the ovaries (long arrows). Bright T2 signal follicles make the ovaries identifiable in 90% of women of menstrual age.
(C) The normal H or butterfly shape of the vagina (short arrows) and the puborecatalis component of the levator ani (long arrows).
the intermediate T2 signal myometrium (Fig. 1). The

endometrium, composed of high signal glands, may
measure up to 1.5 cm in thickness in women of
menstrual age and 5 mm in postmenopausal women.
The junctional zone has been shown to consist of
compacted myometrial cells with a high nuclear to
cytoplasmic ratio [1]. The junctional zone should
measure no more than 11 mm in thickness [2]. It
may become difficult to identify in elderly women
and those women taking oral contraceptives.
Inferiorly, the junctional zone widens to become
the low T2 signal intensity cervical stroma. Within
the cervical stroma, two other zones can be identified.
There is a central very high signal region consisting
of mucus and glandular tissue. Just adjacent to this is

the frond-like plica palmatae [3]. On sagittal and axial
images the vaginal fornices delineate the external os.
The vagina is an intermediate signal intensity H- or
butterfly-shaped organ that is closed at rest.
In women of menstrual age, the ovaries should be
identified in virtually every case. They are ovoid
structures containing high signal follicles. In
postmenopausal women the ovaries are atrophied
and identified in only 40% of cases. The bladder
usually contains high signal urine and has a low T2
signal muscular wall. The urethra is a round structure
with a target appearance. The levator ani and muscles
of the pelvic sidewall are of intermediate signal inten-
sity. Fascial condensations supporting the pelvic

organs are usually not visible.
Benign conditions
Mullerian anomalies
The mullerian ducts are paired structures that fuse
between weeks 6 and 11 of gestation. They form the
uterus, cervix, fallopian tubes, and upper two thirds
of the vagina. The lower one third of the vagina arises
from the urogenital sinus. Anomalies occur in less
than 5% of the general population but are present in
approximately 20% of those patients who present
with multiple spontaneous first-trimester abortions
[4]. Anomalies occur because of failure of development or fusion of the mullerian ducts and failure of
septal resorption once fusion has occurred. Although
often initially diagnosed using HSG, MR imaging
should be performed when surgical therapy is considered to characterize more accurately the anomaly.
It is particularly important to differentiate a septate
from a bicornuate uterus, each of which requires
different therapy.
The most common mullerian anomalies are caused
by in utero exposure to diethylstilbestrol. This medication, intended to thwart premature labor, was last
given in 1970. It is estimated that 60% of exposed
individuals have an abnormal HSG. Abnormal findings include a squared off uterine contour (caused by
myometrial hypertrophy), uterine hypoplasia, and
fallopian tube diverticula [5,6]. The diagnosis of
181
diethylstilbestrol exposure is almost always made

through a combination of careful medical history
taking and HSG. MR imaging is rarely required.
Failure of development of the mullerian ducts may
be partial or complete and unilateral or bilateral.
Complete failure of development results in uterine
agenesis. This finding in association with congenital
anomalies, usually of the skeleton, comprises Rokitansky-Kuster-Hauser syndrome. In these patients,
the external vagina has a normal appearance.
Failure of development of one duct leads to
unicornuate uterus (Fig. 2). This anomaly is identified as a small, lateroflexed uterine cavity on both
HSG and MR imaging. Once diagnosed, MR images
should be thoroughly scrutinized for the presence of
a rudimentary horn. This horn may be separate from
or connected to the contralateral horn and may
contain endometrial tissue. The presence of endometrial tissue in an isolated rudimentary horn usually
leads to chronic pelvic pain with symptoms similar
to endometriosis.
Complete failure of fusion of the mullerian ducts
yields uterus didelphys (Fig. 3). Two widely separated uterine cavities are identified, each with its own
cervix and cephalad portion of the vagina. In the
minority of cases, a transverse septum obstructs one
of the uterine canals, leading to hematometrocolpos.
Again, the external vagina is of normal appearance.
Partial failure of mullerian duct fusion forms the
bicornuate uterus. This anomaly is characterized by a
concave external fundal contour and two uterine
canals divided by a septum. The septum may be
composed of a mixture of fibrous tissue and myome-
Fig. 2. Coronal T2-weighted MR image (A,B) of a didelphys uterus. The anterior image (A) shows the two widely separated
uterine horns, each with its own myometrium, junctional zone, and endometrial canal (arrows). The posterior image (B) shows
two cervices (curved arrow) extending inferiorly to the level of the vaginal fornices.
182
Fig. 3. Axial (A) and coronal (B) T2-weighted MR image of a septate uterus. Axial image (A) shows the characteristic flat
external uterine contour (arrowheads) of the septate uterus. The septum is demarcated with an open arrow. Coronal image (B)
shows a black, fibrous septum extending through the cervices (open arrows) and into the proximal vaginas. The vaginal fornices
are marked with long arrows.
trium. In most cases, a single cervix and vagina are

present. This anomaly, when treated, usually requires
an open metroplasty.
Minimal failure of fusion leads to the arcuate
uterus with a U-shaped internal contour of the fundal
portion of the endometrial canal. This is an entirely
benign variant and has no effect on fertility.
Once fusion of the mullerian ducts has occurred,
the septum resorbs from its caudal-most extent within
the proximal vagina cephalad. Failure of septal
resorption is the cause of the septate uterus (Fig. 4).
In contradistinction to the bicornuate uterus, the
external contour is flat or minimally ( < 1 cm) concave. The septum may be of variable length and
contain fenestrations. Again, the septum often contains a mixture of low T2 signal fibrous and intermediate T2 signal myometrial tissue; however, in
most cases the caudal-most aspect is of homogeneous
low T2 signal indicating a fibrous nature. This
anomaly is often treated with hysteroscopic metroplasty, a day-surgery procedure.
Failure of development of the urogenital sinus
leads to an absence of the caudal two thirds of the
vagina and an obstructed cervix and uterus. The
differential diagnosis is imperforate hymen. On axial
T2-weighted MR imaging of the pelvic floor, the
vagina is absent. A small low T2 signal scar may be
present. Hematometrocolpos manifests as a heterogeneous, but predominantly high T2 signal mass
filling the uterine canal and markedly thinning the
myometrium. Treatment in the young woman often
involves the formation of a neovagina.
Fibroids
Fibroids, or leiomyomata, are present in an estimated 40% of women over 40 years of age and are
particularly common in the black population [7].
Symptoms associated with fibroids include bleeding,
pain, and urinary incontinence. Their presence can
also contribute to infertility. US, either transabdominal or transvaginal, in combination with physical examination, is most often used to diagnose
the presence and monitor the growth of fibroids.
MR imaging is of value in the symptomatic patient
when uterine salvage therapy is considered [8,9].
Techniques include myomectomy, focal endometrial
curettage, hormone administration, and uterine artery embolization.
Using axial and sagittal T2-weighted imaging,
most fibroids, including pedunculated ones, can be
identified with confidence. They should be described
in terms of size, location, and signal intensity. Any
fibroid that impresses on the endometrial canal is
considered to have a submucosal component and can
be a source of bleeding.
Based on T2 signal, fibroids can be classified
further into groups depending on their degree of
cellularity (Fig. 5) [10]. The most common group is
the ordinary fibroids, of relatively homogeneous low
T2 signal and composed of collagenous material.
Cellular fibroids contain less collagen and are of
intermediate T2 signal intensity. These fibroids
respond well to hormonal therapy. Degenerating fibroids are of very bright T2 signal intensity but often
183
Fig. 4. (A) Sagittal and (B) axial T2-weighted MR image of a woman with an enlarged fibroid uterus. A cellular fibroid (black
arrowhead ) of intermediate signal intensity distorts the uterine canal (open arrow).
contain thick septation or wall nodules. When large or

pedunculated they can be difficult to differentiate from
an ovarian neoplasm. Degenerating fibroids do not
respond well to uterine artery embolization because of
central necrosis [11].
hemosiderin rim. Treatment options have traditionally

been limited to hysterectomy. Recently, however,
there has been some reported success with uterine
artery embolization [16].
Endometriosis
Adenomyosis
Adenomyosis is a common cause of pelvic pain in
women of menstrual age. It is defined as extension of
endometrial glandular tissue more than one third of the
depth of the myometrium with adjacent muscular
hypertrophy. Adenomyosis cannot be diagnosed
using CT. With meticulous US technique, it appears
as heterogeneous and cystic areas [12,13]. Because
of its accuracy and ease, MR imaging remains
the diagnostic test of choice (Fig. 6) [14,15]. On T2weighted MR imaging, adenomyosis appears as
poorly defined low T2 signal intensity confluent with
and widening the junctional zone ( > 11 mm) [42].
High T2 signal intensity glands are often seen within
the diseased area. Occasionally, focal adenomyosis
forms a cavity with a very low T2 signal intensity
Laparoscopy remains the test of choice for diagnosing and staging endometriosis. Transvaginal US
and MR imaging are useful only when the diagnosis
of an endometrioma is suspected. Endometriomas
may be unilocular or multilocular and are predominantly high signal on T1- and low or mixed signal
on T2-weighted images [17]. On T2-weighted
images, intermediate signal shading, caused by T2
shortening of blood products, is often seen within the
mass [18]. Recurrent bleeding may lead to a very low
signal hemosiderin rim. Adjacent bowel loops may be
distorted and tethered to the mass. Confident diagnosis can be difficult when some of the previously
mentioned MR imaging features are absent [19].
Differential diagnosis includes dermoid and hemorrhagic cyst.
184
Fig. 5. Sagittal and axial (A,B) T2-weighted MR image of a woman with abdominal pain and an enlarged uterus caused by
adenomyosis. On the sagittal image (A), the abnormal low T2 signal is confluent with and widens the junctional zone nearly to
the edge of the myometrium (arrowheads). The distorted endometrial canal is marked with a star. On the axial image (B), a
dilated endometrial gland (white arrow) extends into the adenomyoma.
Polycystic ovarian disease

The triad of hirsutism, obesity, and amenorrhea
comprises Stein-Leventhal syndrome. Although the
biochemical underpinnings of this syndrome are vari-
able, most patients have elevated levels of luteinizing

hormone. The classic US and MR imaging appearance
of a polycystic ovary is increased central stroma with
peripheral location of follicles (Fig. 7) [20]. The ovary
is enlarged (> 4 cm maximal diameter) in one third of
Fig. 6. Axial T1- (A) and T2-weighted (B) MR images of a woman with chronic pelvic pain caused by endometriosis. The
encapsulated structure in the cul-de-sac is of mixed low and high signal on both T1- and T2-weighted images indicating the
presence of blood products of variable age and diagnostic of an endometrioma (arrows).
Fig. 7. Axial T2-weighted MR image of a woman with

amenorrhea. The ovaries are nearly as large as the uterus and
contain abundant follicles (curved arrows). In this patient
with Stein-Leventhal syndrome, the peripheral pattern of
ovarian follicles was not present.
185
level is present. The Rokitansky nodule, composed of

hair and other components, usually floats within the
center of the mass. On MR imaging, fat can be
identified using fat-suppression pulse sequences
[19]. The chemical shift artifact consisting of an
adjacent black and white line at the edge of the
teratoma in the frequency encoding direction is also
diagnostic of macroscopic fat. On T2-weighted
images, the speckling artifact is diagnostic. It is
composed of myriad chemical shift artifacts occurring
within the Rokitansky nodule. In some cases, dermoid mimics the adjacent bowel and the sigmoid
colon must be reviewed carefully to separate it from
the adnexa. Dermoids without macroscopic fat have
been reported, are rare, and cannot be diagnosed
confidently using any modality [21]. Most asymptomatic dermoids under 4 cm in size are left in place.
Those over 4 cm are often removed because of the
higher risk of ovarian torsion.
Tubo-ovarian complex
cases. There is a strong association with endometrial

hyperplasia and these patients are at increased risk for
development of endometrial carcinoma.
Dermoid
Mature teratoma or dermoid is an ovarian mass
composed of tissue arising from the endoderm, mesoderm, and ectoderm (Fig. 8). It is a benign lesion that
is bilateral in approximately 25% of cases. Identification is usually straightforward. An echogenic mass
with dense posterior shadowing (the tip of the iceberg
sign) is diagnostic on US. On CT, a fat-fat or fat-fluid
Usually a sequelae of pelvic inflammatory disease

or gynecologic instrumentation, tubo-ovarian complex is secondary to infection with gonorrhea or
chlamydia. The fallopian tube becomes dilated and
obstructed, usually by a combination of blood and
pus. The ovary is often involved in the resultant
inflammatory mass, called a tubo-ovarian complex.
The process may be unilateral or bilateral. On contrast-enhanced CT and MR imaging, tubo-ovarian
complex has the appearance of an enhancing mass
with cystic and solid components (Fig. 9) [22]. There
is often extensive stranding of adjacent fat. Differ-
Fig. 8. Axial contrast-enhanced CT (A) and T2-weighted MR images (B) of a woman with a left adnexal dermoid. (A) A large
pelvic mass composed of fat and soft tissue elements (arrows). (B) Obtained at the same level, image shows chemical shift
artifact in the frequency-encoding direction (arrowheads) and the Rokitansky nodule (star) with speckling artifact caused by
myriad sites of chemical shift.
186
Fig. 9. Axial contrast-enhanced CT (A) and T1-weighted, fat-suppressed, contrast-enhanced MR images (B) of a woman with a
right lower quadrant mass. Both images show a primarily cystic mass with a thick enhancing wall and septum (arrow). Two
smaller cystic masses are seen in the left lower quadrant (arrowheads) (A). The differential diagnosis includes ovarian
cystadenocarcinoma. In this case the patient had a long history of pelvic inflammatory disease and sepsis and surgical exploration
revealed bilateral tubo-ovarian abscesses.
ential diagnosis includes torsion, diverticulitis, appendicitis, and ovarian neoplasm; however, the presence
of fever, cervical motion tenderness, and appropriate
history usually points to the correct diagnosis. Primary treatment is antibiotic therapy for 48 to
72 hours. If this regimen fails, percutaneous or
surgical drainage is usually performed.
Gynecologic neoplasms
Most gynecologic oncologists stage pelvic neoplasms according to the FIGO system (International
Federation of Gynecology and Obstetrics). This form
of clinical staging relies on a high-quality physical
examination performed under general anesthesia.
Numerous radiologic tests are then performed including chest radiograph, barium enema, and intravenous
pyelogram. It has been reported by two groups of
investigators that performing a single MR exam of
the pelvis can replace all of the ancillary examinations with the exception of the chest radiograph at a
slightly lower cost [23,24].
Some gynecologic oncologists argue that preoperative imaging is of little value because once a
significant gynecologic mass is detected, it must be
removed using an extensive oncologic-type resection
that is both diagnostic and therapeutic. There is
agreement that CT or MR imaging is of particular
value in the case of very advanced disease, where the
primary therapy should be chemotherapy and not
surgery, and to precede or replace the second-look

operation in identification of recurrent metastases.
MR imaging should consist of sagittal and axial
T2-weighted images, optimally obtained with a multicoil array and fast-pulse sequences and dynamic
images immediately postintravenous administration
of gadolinium pentetic acid (Gd-DTPA). These contrast-enhanced images better characterize cervical and
ovarian masses and can separate them from adjacent
structures, such as the bladder and rectum.
Cervical cancer
Cervical cancer is usually a disease of women of
menstrual age that plateaus at age 40 in American
women. Approximately 13,000 new cases are diagnosed each year [25]. Causality by human papilloma
virus generally has been accepted and most tumors
are of squamous cell origin. Patients are usually
asymptomatic and detection is usually by the Papanicolaou smear. Most cancers are confined to the
cervical canal when discovered. They are treated with
a variety of local therapies including laser ablation.
Any cervical mass that is greater than 1.5 cm or
presumed to extend beyond the cervix should be
examined using MR imaging. CT scanning lacks
the contrast resolution necessary for primary staging
of relatively low-volume disease.
On T2-weighted MR imaging, cervical cancer is
usually hyperintense to the dark cervical stroma. The
preservation of the black ring of the cervical stroma,
no matter how thin, virtually excludes parametrial

extension. These patients are candidates for surgical
cure (stage IIA). Those in whom the black line is
broken and a mass extends beyond the expected
confines of the cervix are usually treated primarily
with brachytherapy (stage IIB). In cases of advanced
disease, sagittal T2-weighted or Gd-DTPA enhanced
T1-weighted images show invasion of the rectum,
bladder, and vaginal fornices (Fig. 10) [2,26]. Lymph
nodes greater than 1 cm in size are positive for tumor
in 65% of cases. Several small ( < 1 cm) lymph nodes
in a chain are also suspicious for extension of disease.
In cases of advanced disease, CT is preferred. It is
quick to perform and read and has higher spatial
resolution than MR imaging. The cervix is often
enlarged in the anteroposterior direction and may
contain gas. Images should be scrutinized for extension into the pelvic sidewall and enlargement of pelvic
and retroperitoneal lymph nodes (Fig. 11). Extension
of disease is also manifested by thickening of the
uterosacral ligaments, although this can also be a
result of radiation. The development of hydronephrosis is often caused by bladder wall or ureteric invasion
and upstages the disease to IIIB. Finally, using CT
scans, metastases to other visceral organs can be
detected without requiring extra scanning time.
187
Endometrial cancer
Endometrial cancer is the most common of the
gynecologic malignancies. It also has the best prognosis. Of the 38,000 new cases diagnosed each year,
75% of patients present with stage I disease confined
to the endometrial canal and curable by hysterectomy
[25]. Presentation is usually painless vaginal bleeding
in a postmenopausal woman and detection is usually
done by a combination of US and pipette biopsy. Risk
factors include increased age and prolonged exposure
to unopposed estrogens.
The endometrial thickness of postmenopausal
women should not exceed 5 mm. Women with
thickened endometrial linings on US examination
merit biopsy. Even in the presence of vaginal
bleeding, however, endometrial thickness of less
than 5 mm on US excludes endometrial cancer.
Women taking hormone replacement therapy or
tamoxifen may have slightly thicker endometrial
linings (8 mm); however, 5 mm should still be
taken as the upper limits of normal because these
women are at a higher risk for endometrial carcinoma [27 29].
Most women with endometrial carcinoma do not
require imaging. Treatment is predicated on physical
Fig. 10. Sagittal (A) and axial (B) T2-weighted MR images of a woman with a large cervical cancer. On the sagittal image (A)
the tumor bulk is delineated by arrows. The anterior aspect of the tumor invades the bladder wall. (B) A markedly enlarged
cervix with tumor replacing the entirety of the normal low T2 signal stroma (arrowheads). This finding alone has an 85%
positive predictive value for extension of tumor beyond the confines of the cervix and into the parametria.
188
Fig. 11. Contrast-enhanced axial CT images of the abdomen and pelvis of a woman with advanced cervical cancer. (A) The
cervix is enlarged and contains a large amount of gas, likely caused by tumor necrosis (short arrows). A large lymph node abuts
the right pelvic sidewall (long arrow). (B) There is right hydronephrosis (arrowhead) and hydroureter, likely caused by invasion
and occlusion of the ureterovesical junction.
examination and low-stage disease confined to

the endometrium. MR imaging is recommended
when locally advanced disease is expected based
on physical examination findings and in the patient
with a difficult physical examination because of
obesity, prior radiation, or surgery [30]. Sagittal and
axial T2-weighted images and contrast-enhanced
images should be performed and scrutinized for
extension of hyperintense tumor into or through the
myometrium (Fig. 12). Preservation of a subendometrial band of enhancement virtually excludes myometrial extension [1]. Invasion of greater than 50% of
the myometrial wall thickness is strongly indicative
of involved lymph nodes and a poorer overall prognosis [31,32]. In the minority of patients, endometrial carcinoma presents as an obstructing lower
uterine segment mass with associated hematometra
or pyometra. Metastases can extend by the fallopian
tubes to the ovaries causing development of an
adnexal mass. Advanced disease is probably again
best assessed using contrast-enhanced CT scan.
Ovarian cancer
Ovarian cancer is a disease of perimenopausal
women with its peak incidence between the ages of
45 and 55 years. Approximately 23,000 new cases are
identified each year [25]. Risk factors include a family
history of ovarian carcinoma and, in some cases,
breast cancer. Most tumors are of epithelial origin
and are serous or mucinous cystadenocarcinomas.
Unfortunately, no effective screening examination
exists and most women (60%) present with advanced
disease extending beyond the pelvis. A recent radiologic diagnostic oncologic trial showed that US, CT,
and MR imaging are equal in the detection and overall
staging of ovarian cancer [33]. MR imaging has a

slightly higher detection rate for peritoneal metastases
[34]. In routine practice, CT is usually preferred
because it is well accepted by the patient and provides
a rapid overall view of the abdomen and pelvis that is
well understood by referring physicians. MR imaging,
however, is particularly useful in the detection of
recurrent disease because its tremendous contrast
sensitivity yields excellent identification of even very
small cystic implants.
Cystadenocarcinoma usually appears on both CT
and MR imaging as a predominantly cystic mass with
a nodular, enhancing wall and thick, enhancing septa.
Ascites, hydronephrosis, and omental and peritoneal
implants are common accompaniments. Enlarged
lymph nodes are less common but do occur. Cystadenocarcinoma with ascites and minimal implant
formation is treated with an extensive surgery. The
procedure is comprised of hysterectomy, bilateral
salpingo-oophorectomy, lymphadenectomy, ascites
and peritoneal sampling, omentectomy, and manual
examination of the surface of the liver. On imaging
examinations it is important to identify solid implants
before therapy because a large solid tumor burden is
best treated with chemotherapy rather than primary
debulking surgery.
Pelvic floor relaxation

Approximately 50% of parous women in the
United States have pelvic floor relaxation and 20%
of that group is symptomatic enough to seek treatment. Symptoms include urinary and fecal incontinence and protrusion of tissue, usually the cervix or
uterus, through the pelvic floor. Pelvic floor relaxa-
189
Fig. 12. Sagittal and axial T2-weighted (A,B) and axial contrast-enhanced T1-weighted (C) MR images of the uterus of a woman
with painless vaginal bleeding. (A,B) An intermediate signal mass within and widening the endometrial canal (arrowheads). (B)
A suggestion of endometrial invasion of the anterior aspect of the myometrium. This finding is confirmed on the contrastenhanced image (C). The endometrial cancer extends less than 50% of myometrial thickness indicating a good long-term
prognosis (arrow).
tion is usually caused by a combination of muscle

damage and fascial stretching or tearing sustained
during childbirth. Most women with urinary incontinence can be diagnosed and treated based on
physical examination findings and office urodynamics. Pelvic floor imaging using MR imaging is
indicated when multiple compartments of the pelvic
floor are involved and when planning repeat surgeries. Sagittal midline images using an ultrafast T2weighted pulse sequence with the woman at rest and
at maximal strain quantify the descent of all three
compartments at once and can be used to identify
enterocele, sigmoidocele, and anterior urethral rotation and kinking. Descent of organs 2 cm below the
pubococcygeal line usually indicates the need for
surgery in a symptomatic patient [35]. Thin, axial,

high-resolution T2-weighted images detail muscle
atrophy and tears [36]. Lateral deviation of the vagina
usually indicates a paravaginal fascial tear.
Pregnancy
Imaging of the pregnant mother or fetus is done
only in certain circumstances and following a clear
discussion among the patient, referring physician, and
radiologist of possible risks and benefits. In the case of
significant maternal trauma, contrast-enhanced CT
scan is the test of choice. It can be done quickly and
is extremely reliable in the detection of hemoperito-
190
neum and visceral injury. The most common cause of

fetal death is maternal death, followed by placental
abruption. All scan parameters should be recorded and
given to the departmental physicist who then makes an
estimate of maternal dose. In general, a dose of greater
than 5 rad should generate a consideration of abortion
in a first-trimester pregnancy.
MR imaging is often requested for evaluation of
maternal pathology ranging from headaches and back
pain to cancer staging. Although no teratogenic effects
have been reported secondary to pulse sequences
routinely used in human imaging, it is still prudent
to discuss the possible maternal benefits and the
unknown long-term risk of MR imaging to the fetus
[37,38]. One of the more common requests is the
characterization of a maternal pelvic mass identified
on US examination (Fig. 13). In most cases, imaging
should be done in the second trimester because the
corpus luteum cyst has resolved and it is the optimal
time for surgery. All scans should be monitored by the
radiologist. Pulse sequences should be kept to a
minimum and Gd-DTPA administered only when
necessary. It is critically important that the origin of
any pelvic mass be identified [39]. A pedunculated
fibroid is left in place, whereas a presumed ovarian or
other malignancy is operatively removed.
CT and MR imaging are both useful in the

peripartum state. Flow-sensitive gradient echo MR
imaging pulse sequences through the pelvis are a
particularly fast and accurate way to exclude pelvic or
ovarian vein thrombosis [40]. CT scan is usually the
test of choice for identification of abscess secondary
to endometritis or wound infection.
Fetal anomalies
With the advent of ultrafast T2-weighted pulse
sequences, such as half Fourier single-shot turbo spin
echo and SSFSE, it has become possible to image the
fetus without the blurring associated with motion. An
MR image is first obtained parallel to the fetal spine.
Each set of images then serves as the scout for the
next, including fetal axial and coronal planes [41].
Although US remains the mainstay in the detection of
fetal anomalies, MR imaging has proved of value in
confirming visceral and musculoskeletal anomalies
and identifying cranial anomalies. Imaging is usually
done as part of a complete clinical work-up in a parent
known to carry a second-trimester fetus with an
abnormal karyotype or US. The most common indication is myelomeningocele. Findings can be used to
Fig. 13. Transabdominal ultrasound (A) and sagittal T2-weighted MR images (B) of a pregnant woman with pelvic pain
following cerclage. (A) The fetal head is demarcated by a white arrow. A fluid collection is seen within the cul-de-sac. (B) The
fluid collection is marked with a star and the gravid uterus with long arrows. Both ovaries were identified on other MR images
(not shown) eliminating the possibility of a hydrosalpinx or cystadenoma. Percutaneous drainage was curative and demonstrated
abundant white blood cells indicative of abscess.
plan for in utero surgery or to counsel parents on

prognosis and the need for further genetic testing.
[11]
Summary
MR imaging has become a valuable modality in
the evaluation of the female pelvis. In many cases, it
follows the performance of HSG or US. These cases
include infertility and pelvic pain. In some cases, it
should serve as the test of choice. This is true in the
local staging of cervical cancer and the evaluation of
pain or disability in the pregnant patient. Finally, in
the evaluation of advanced gynecologic cancers, it is
usually a secondary choice with CT preferred.
References
[1] Seki H, Takano T, Sakai K, et al. Value of dynamic MR
imaging in assessing endometrial carcinoma involvement of the cervix. AJR Am J Roentgenol 2000; 175:
171 6.
[2] Scheidler J, Heuck AF, Steinborn M, et al. Parametrial
invasion in cervical carcinoma: evaluation of detection
at MR imaging with fat suppression. Radiology 1998;
206:125 9.
[3] Scoutt LM, Flynn SD, Luthringer DJ, et al. Junctional
zone of the uterus: correlation of MR imaging and
histologic examination of hysterectomy specimens.
Radiology 1991;179:403 7.
[4] Sorensen SS. Estimated prevalence of mullerian
anomalies. Acta Obstet Gynecol Scand 1998;67:
441 5.
[5] Van Gils APG, Tham RTOTA, Falke THM, Peters
AAW. Abnormalities of the uterus and cervix after
diethylstilbestrol exposure: correlation of findings on
MR and hysterosalpingography. AJR Am J Roentgenol
1989;153:1235 8.
[6] Kaufman RH, Noller K, Adam E, et al. Upper genital
tract abnormalities and pregnancy outcome in diethylstilbestrol exposed progeny. Am J Obstet Gynecol
1984;148:973 84.
[7] Creasman WT. Disorders of the uterine corpus. In:
Scoutt JR, DiSaia PJ, Hammand CCB, Spellacy WN,
editors. Danforths obstetrics and gynecology. Philadelphia: Lippincott; 1994. p. 925 55.
[8] Murase E, Siegelman ES, Outwater EK, et al. Uterine
leiomyomas: histopathologic features, MR imaging
findings, differential diagnosis, and treatment. Radiographics 1999;19:1179 97.
[9] Dudiak CM, Turner DA, Patel DK, et al. Uterine
leiomyomas in the infertile patient: preoperative localization with MR imaging versus US and hysterosalpingography. Radiology 1988;167:627 60.
[10] Yamashita Y, Torashima M, Takahashi M, et al. Hyperintense uterine leiomyoma at T2-weighted MR imag-
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
191
ing and clinical implications. Radiology 1993;189:

721 5.
Burn PR, McCall JM, Chinn RJ, et al. Uterine fibroleiomyoma: MR imaging appearances before and after
embolization of uterine arteries. Radiology 2000;
214:729 34.
Reinhold C, Mostafa A, Mehio A, et al. Diffuse uterine
adenomyosis: morphologic criteria and diagnostic accuracy of endovaginal sonography. Radiology 1995;
197:609 14.
Atri M, Reinhold C, Mehio A, et al. Adenomyosis: US
features with histologic correlation in an in vitro study.
Radiology 2000;215:783 90.
Ascher SM, Arnold LL, Patt RH, et al. Adenomyosis:
prospective comparison of MR imaging and transvaginal sonography. Radiology 1994;190:803 6.
Mark AS, Hricak H, Heinrichs LW, et al. Adenomyosis
and leiomyoma: differential diagnosis with MR imaging. Radiology 1987;163:527 9.
Siskin GP, Tublin ME, Stainken BF, et al. Uterine
artery embolization for the treatment of adenomyosis:
clinical response and evaluation with MR imaging.
Gougoutas CA, Siegelman ES, Hunt J, et al. Pelvic
endometriosis: various manifestations and MR imaging
findings. AJR Am J Roentgenol 2000;175:353 8.
Outwater E, Schiebler ML, Owen RS, et al. Characterization of hemorrhagic adnexal lesions with MR imaging: blinded reader study. Radiology 1993;186:489 94.
Yamashita Y, Torashima M, Hatanaka Y, et al. Adnexal
masses: accuracy of characterization with transvaginal
US and precontrast and postcontrast MR imaging. Radiology 1995;194:557 65.
Kimura I, Togashi K, Kawakami S, et al. Polycystic
ovaries: implications of diagnosis with MR imaging.
Radiology 1996;201:549 52.
Yamashita Y, Hatanaka Y, Torashima M, et al. Mature
cystic teratomas of the ovary without fat in the cystic
cavity: MR features in 12 cases. AJR Am J Roentgenol
1994;163:613 6.
Tukeva TA, Aronen HJ, Karjalainen PT, et al. MR imaging in pelvic inflammatory disease: comparison with
laparoscopy and US. Radiology 1999;210:209 16.
Hardesty LA, Sumkin JH, Nath ME, et al. Use of preoperative MR imaging in the management of endometrial carcinoma: cost analysis. Radiology 2000;215:
45 9.
Kinkel K, Kagi Y, Yu K, et al. Radiologic staging in
patients with endometrial cancer: a meta-analysis. Radiology 1999;212:711 8.
Ahlmedin J, Thomas A, Murray T, Thun M. Cancer
statistics, 2002. CA Cancer J Clin 2002;52:23 45.
Hawighorst H, Knapstein PG, Weikel W, et al. Cervical
carcinoma: comparison of standard and pharmacokinetic MR imaging. Radiology 1996;201:531 9.
Levine D, Gosink BB, Johnson LA. Change in endometrial thickness in postmenopausal women undergoing hormone replacement therapy. Radiology 1995;
197:603 8.
192
[28] Hann LE, Giess CS, Bach AM, et al. Endometrial

thickness in tamoxifen-treated patients: correlation
with clinical and pathologic findings. AJR Am J
Roentgenol 1997;168:657 61.
[29] Ascher SM, Johnson JC, Barnes WA, et al. MR imaging appearance of the uterus in postmenopausal women receiving tamoxifen therapy for breast cancer:
histopathologic correlation. Radiology 1996;200:
105 10.
[30] Hricak H, Rubinstein LV, Gherman GM, et al. MR
imaging evaluation of endometrial carcinoma: results
of an NCI cooperative Study. Radiology 1991;179:
829 32.
[31] Boronow RC, Morrow CP, Creasman WT, et al. Surgical staging in endometrial cancer: clinical-pathologic
findings of a prospective study. Obstet Gynecol 1984;
63:825 32.
[32] Frei KA, Kinkel K, Bonel HM, et al. Prediction of
deep myometrial invasion in patients with endometrial
cancer: clinical utility of contrast-enhanced MR imaging-a meta-analysis and bayesian analysis. Radiology
2000;216:444 9.
[33] Kurtz AB, Tsimikas JV, Tempany CM, et al. Diagnosis
and staging of ovarian cancer: comparative values of
Doppler and conventional US, CT, and MR imaging
correlated with surgery and histopathologic analysisreport of the radiology diagnostic oncology group. Radiology 1999;212:19 27.
[34] Low RN, Saleh F, Song SY, et al. Treated ovarian
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
cancer: comparison of MR imaging with serum

CA-125 level and physical examination: a longitudinal study. Radiology 1999;211:519 28.
Hoyte L, Schierlitz L, Zou K, et al. Two- and 3-dimensional MRI comparison of levator ani structure, volume, and integrity in women with stress incontinence
and prolapse. Obstet Gynecol 2001;185:11 9.
Fielding JR, Griffiths DJ, Versi E, et al. MR imaging of
pelvic floor continence mechanisms in the supine and
sitting positions. AJR Am J Roentgenol 1998;171:
1607 10.
Consensus Conference. Magnetic resonance imaging.
JAMA 1998;259:2132 8.
Shellock FG, Litwer CA, Kanal E. Magnetic resonance
imaging: bioeffects, safety, and patient management.
Kier B, McCarthy SM, Scoutt LM, et al. Pelvic
masses in pregnancy: MR imaging. Radiology
1990;176:709 13.
Twickler DM, Setiawan AT, Evans RS, et al. Imaging
of puerperal septic thrombophlebitis: prospective comparison of MR imaging, CT, and sonography. AJR Am
J Roentgenol 1997;169:1039 43.
Levine D, Barnes PD, Edelman RR. Obstetric MR
imaging. Radiology 1999;211:609 17.
Kang S, Turner DA, Foster GS, Rapoport MI, et al.
Adenomyosis: specificity of 5mm as the maximum
normal uterine junctional zone thickness in MR images. AJR Am J Roentgenol 1996;166:1145 50.
Radiol Clin N Am 41 (2003) 193 197
Index
Note: Page numbers of article titles are in boldface type.
Arterial obstruction, hepatic, MR imaging of, 85
Abdominal aorta, MR angiography of, 124 131, 133

for aneurysms, 127 129
for aortic dissection, 129
for aortic occlusion, 129 130
for renal artery stenosis, 130
for renal transplant evaluation, 130 131, 133
Atherosclerotic plaque, MR imaging of, 21
Adenomas, adrenal, MR imaging of, 153 154

hepatocellular, MR imaging of, 59 60
serous microcystic, MR cholangiopancreatography
of, 107
Adenomyosis, MR imaging of, 183
Adrenal glands, MR imaging of, 145 146, 152 157
for adenomas, 153 154
for adrenocortical carcinoma, 154 155
for cysts, 156 157
for myelolipomas, 156
for pheochromocytomas, 155 156
for pseudocysts, 156 157
technique for, 145 146
Adrenocortical carcinoma, MR imaging of, 154 155
Alcoholic pancreatitis, MR cholangiopancreatography
of, 102 103
Aneurysms, abdominal aortic, MR angiography of,
127 129
thoracic aortic, MR angiography of, 35 36
Angiomyolipomas, MR imaging of, 148 150
Aortic aneurysms, abdominal, MR angiography
of, 127 129
thoracic, MR angiography of, 35 36
B
Bicornuate uterus, MR imaging of, 181 182
Biliary complications, postoperative, MR cholangiopancreatography of, 92 93
Biliary cystadenomas, MR imaging of, 60
Biliary cysts, MR cholangiopancreatography of,
90 91
Bladder, MR imaging of, 161 177
anatomy in, 162
for cystoceles, 173 175
for diverticula, 171 172
for endometriosis, 172 173
for lymphomas, 170
for mesenchymal tumors, 170 171
for tumors, 162 170
contrast-enhanced, 167 168
direct spread, 165 167
epidemiology of, 162 163
management and follow-up, 169 170
metastatic to bones, 168 169
metastatic to lymph nodes, 169
primary mass, 163 165
for vesical congestion and inflammation, 172
patient preparation for, 161
planes in, 161 162
pulse sequences in, 161 162
Blood pool agents, in MR angiography, 140
Aortic dissection, abdominal, MR angiography

of, 129
thoracic, MR angiography of, 34 35
Bolus-chase MR angiography, technique for,

134 136, 138
Aortic occlusion, abdominal, MR angiography of,

129 130
Breasts, MR imaging of, 43 50

for fibroadenomas, 47 48
image analysis in, 45 49
Arrhythmogenic right ventricular dysplasia,

MR imaging of, 21 22
Breast cancer, screening for, MR imaging in, 49 50
PII: S 0 0 3 3 - 8 3 8 9 ( 0 3 ) 0 0 0 0 3 - 4
194
Index / Radiol Clin N Am 41 (2003) 193197
patient preparation for, 44

techniques for, 44
to screen for breast cancer, 49 50
Cholecystolithiasis, MR cholangiopancreatography
of, 91
Budd-Chiari syndrome, MR imaging of, 84 85
Choledocholithiasis, MR cholangiopancreatography
of, 91 92
Bypass graft patency, MR imaging of, 20 21
Cirrhosis, MR imaging of, 68 74

Congenital heart disease, MR imaging of, 25
Congestive heart failure, hepatic efffects of, 85
Cardiac masses, MR imaging of, 24
Contrast enhancement, of MR angiography,

of abdominal aorta, 116, 118 131, 133
of chest, 29 31
of MR imaging, of bladder tumors, 167 168
of breasts, 44 45
of cardiovascular disease, 18 19
of hepatic tumors. See Hepatic tumors.
Cardiomyopathies, MR imaging of.

See Cardiovascular disease.
Cardiovascular system, MR imaging of, 17 28
for cardiac masses, 24
for cardiomyopathies, 21 23
arrhythmogenic right ventricular dysplasia,
21 22
dilated, 23
hemochromatosis, 23
hypertrophic, 22
restrictive, 23
for congenital heart disease, 25
for coronary arteries, 19 21
anomalous arteries, 20
atherosclerotic plaque, 21
bypass graft patency, 20 21
Kawasaki disease, 20
MR angiography in, 19 20
for ischemic heart disease, 17 19
myocardial viability in, 17
rest imaging, 17
stress imaging, 17 18
for pericardial disease, 23 24
for valvular heart disease, 21
Central thoracic veins, MR angiography of, 37
Cervical cancer, MR imaging of, 186 187
Chest, MR angiography of, 29 41
for aortic aneurysms, 35 36
for aortic dissection, 34 35
for central thoracic veins, 37
for developmental anomalies, 36
for pulmonary arteries, 37 39
for thoracic aorta, 34
image analysis in, 34
techniques for, 29 34
Cholangiocarcinoma, MR cholangiopancreatography
of, 93 94
MR imaging of, 62 63
Coronary arteries, MR angiography of, 19 21

Cystadenocarcinoma, MR imaging of, 188
Cystadenomas, biliary, MR imaging of, 60
Cystoceles, bladder, MR imaging of, 173 175
Cysts, adrenal, MR imaging of, 156 157
biliary, MR cholangiopancreatography of, 90 91
hepatic, MR imaging of, 56 57
renal, MR imaging of, 151 152
D
Dermoids, MR imaging of, 185
Diffuse hepatic disease, MR imaging of, 67 87
arterial obstruction, 85
Budd-Chiari syndrome, 84 85
cirrhosis, 68 74
congestive heart failure, 85
fatty liver, 77 79
hemochromatosis, 23, 79 80
hepatitis, 74 75, 77
hepatocellular carcinoma, 62, 81
lymphomas, 81
metastatic disease, 81
mucopolysaccharidoses, 80 81
portal vein thrombosis, 81, 84
vascular disorders, 81
Dilated cardiomyopathy, MR imaging of, 23
Diverticula, bladder, MR imaging of, 171 172
Dobutamine stress imaging, of cardiovascular
disease, 17 18
Double-duct sign, in MR cholangiopancreatography, 105
Duct-penetrating sign, in MR cholangiopancreatography, 105
E
Echo planar MR imaging, physics of, 9 12
Endometrial cancer, MR imaging of, 187 188
Endometriosis, MR imaging of, 172 173, 183
Endoscopic retrograde cholangiopancreatography,
versus MR cholangiopancreatography, 89,
99 100
F
Fast spin echo MR imaging, physics of, 6 7, 9
Fatty liver, MR imaging of, 77 79
Female pelvis, MR imaging of, 179 192
anatomy in, 179 181
for adenomyosis, 183
for cervical cancer, 186 187
for dermoids, 185
for endometrial cancer, 187 188
for endometriosis, 183
for fetal anomalies, 1990 191
for fibroids, 182 183
for mullerian anomalies, 181 182
for ovarian cancer, 188
for pelvic floor relaxation, 188 189
for polycystic ovarian disease, 184 185
for tubo-ovarian complex, 185 186
in pregnancy, 189 190
technique for, 179
Hepatic tumors, MR imaging of, 51 65

biliary cystadenomas, 60
cholangiocarcinoma, 62 63
ferumoxides in, 53 54
gadolinium chelates in, 52 53
hepatocyte-selective, 54 55
perfusional, 55 56
cysts, 56 57
focal nodular hyperplasia, 58 59
hemangiomas, 57 58
hepatocellular adenomas, 59 60
hepatocellular carcinoma, 62, 81
metastatic disease, 60 62, 81
T1-weighted sequences in, 51 52
T2-weighted sequences in, 52
Hepatitis, MR imaging of, 74 75, 77
Hepatocellular adenomas, MR imaging of, 59 60
Hepatocellular carcinoma, MR imaging of, 62, 81
Hepatocyte-selective contrast agents, in MR imaging,
54 56
Hypertrophic cardiomyopathy, MR imaging of, 22
I
Intraductal papillary mucinous tumors, MR cholangiopancreatography of, 107 110
Iron deposition, hepatic, MR imaging of, 79 80
Ischemic heart disease, MR imaging of.
See Cardiovascular system.
Ferumoxides, in MR imaging, 53 54
Fetal anomalies, MR imaging of, 190 191
Kawasaki disease, MR imaging of, 20
Fibroadenomas, of breasts, MR imaging of, 47 48
Kidneys, MR imaging of, 145 152

for angiomyolipomas, 148 150
for cysts, 151 152
for lymphomas, 150
for metastatic disease, 150 151
for preoperative planning, 152
for renal cell carcinoma, 146 148
Fibroids, uterine, MR imaging of, 182 183

Focal nodular hyperplasia, hepatic, MR imaging of,
58 59
H
Half-Fourier acquisition single-shot turbo spin echo
sequence, in MR imaging, 52, 68
Klatskins tumors, MR cholangiopancreatography of,

93 94
Hamartomas, renal, MR imaging of, 148 150

Hemangiomas, hepatic, MR imaging of, 57 58
Hemochromatosis, MR imaging of, 23, 79 80
Hepatic disease, diffuse. See Diffuse hepatic disease.
195
L
Leiomyomas, bladder, MR imaging of, 170
uterine, MR imaging of, 182 183
196
of peripheral arterial occlusive disease,

133 136, 138
bolus chasing in, 134 136, 138
phase-contrast, 115 116
time-of-flight, 115 116
Leriches syndrome, MR angiography of, 129 130

Liver tumors. See Hepatic tumors.
Lung cancer, metastatic to kidneys, MR imaging of,
150 151
Lymph nodes, involvement of, by bladder tumors,
MR imaging of, 169
MR cholangiopancreatography, of pancreatic diseases. See Pancreatic diseases.
Lymphomas, bladder effects of, MR imaging of, 170

hepatic effects of, MR imaging of, 81
renal effects of, MR imaging of, 150
MR urography, technique for, 146

Mucinous cystic neoplasms, MR cholangiopancreatography of, 107
Mucopolysaccharidoses, MR imaging of, 80 81
M
Magnetic resonance imaging, of adrenal glands.
See Adrenal glands.
of bladder. See Bladder.
of breasts. See Breasts.
of cardiovascular system.
See Cardiovascular system.
of diffuse hepatic disease. See Diffuse
hepatic disease.
of female pelvis. See Female pelvis.
of hepatic tumors. See Hepatic tumors.
of kidneys. See Kidneys.
physics of, 1 15
conventional image acquisition, 3
image data, 2 3
motion artifacts, 3
motion compensation approaches to,
35
pulse sequences, 5 7, 9 12, 161 162
echo planar imaging, 9 12
fast spin echo imaging, 6 7, 9
spiral imaging, 9 12
reduction of data collection in, 12 14
requirements and challenges of, 1
Mangafodipir, in MR imaging, 54 55
Mesenchymal tumors, bladder, MR imaging of,
170 171
Metastatic disease, to bones, MR imaging of,
168 169
to kidneys, MR imaging of, 150 151
to liver, MR imaging of, 60 62, 81
MR angiography, contrast-enhanced, 116, 118 124
bolus delivery in, 123 124
timing of, 119 121
future directions in, 138 140
of abdominal aorta. See Abdominal aorta.
of chest. See Chest.
of coronary arteries, 19 21
Mullerian anomalies, MR imaging of, 181 182

Myelolipomas, MR imaging of, 156
Myocardial viability, MR imaging of, 17
N
Neurofibromas, bladder, MR imaging of, 170
O
Ovarian cancer, MR imaging of, 188
P
Pancreas divisum, MR cholangiopancreatography of,
91, 101
Pancreatic diseases, MR cholangiopancreatography
of, 89 96, 97 114
anatomy in, 100 101
biliary congenital variants, 91
biliary cystic diseases, 90 91
cancer, 94
cholangiocarcinoma, 93 94
cholecystolithiasis, 91
choledocholithiasis, 91 92
cysts, 105 110
future directions in, 94 95
pancreas divisum, 91, 101
pancreatitis, 93, 101 104
versus pancreatic cancer, 104 105
postoperative biliary complications, 92 93
primary sclerosing cholangitis, 92
santoriniceles, 101
technique for, 89 90, 97 99
versus endoscopic retrograde cholangiopancreatography, 89, 99 100
Pancreatitis, MR cholangiopancreatography of, 93,
101 104
versus pancreatic cancer, MR cholangiopancreatography of, 104 105

Paramagnetic contrast agents, in MR angiography, of
chest, 30 31
197
Restrictive cardiomyopathy, MR imaging of, 23

Rhabdomyosarcomas, bladder, MR imaging of, 171
Right ventricular dysplasia, arrhythmogenic, MR
imaging of, 21 22
Pelvic floor relaxation, MR imaging of, 188 189

Pelvis, female. See Female pelvis.
Perfusional contrast agents, in MR imaging, 55 56

Pericardial disease, MR imaging of, 23 24
Santoriniceles, MR cholangiopancreatography
of, 101
Peripheral arterial occlusive disease,

MR angiography of, 133 136, 138
Secretin-enhanced dynamic MR cholangiopancreatography, technique for, 99
Phase-contrast MR angiography, technique for,

115 116
Segmented volume acquisition, in MR

angiography, 139
Phase ordering with automatic window selection, in

MR imaging, 5
Serous microcystic adenomas, MR cholangiopancreatography of, 107
Pheochromocytomas, adrenal, MR imaging of,

155 156
bladder, MR imaging of, 170
Spiral MR imaging, physics of, 9 12
Polycystic ovarian disease, MR imaging of, 184 185

Portal hypertension, in cirrhosis, MR imaging of, 74
Portal vein thrombosis, MR imaging of, 81, 84
Steady-state free precession, in MR angiography, 33

Steatosis, MR imaging of, 77 79
Stein-Leventhal syndrome, MR imaging of, 184 185
Stress imaging, of cardiovascular disease, 17 18
Pregnancy, MR imaging in, 189 190

Primary sclerosing cholangitis,
MR cholangiopancreatography of, 92
MR imaging of, 74
Pseudocysts, adrenal, MR imaging of, 156 157
pancreatic, MR cholangiopancreatography of,
105 106
Pseudosarcomas, bladder, MR imaging of, 171
Pseudotumors, bladder, MR imaging of, 171
Pulmonary arteries, MR angiography of, 37 39
T
Teratomas, MR imaging of, 185
Teslascan, in MR imaging, 54 55
Thoracic aorta, MR angiography of, 34
Thoracic veins, central, MR angiography of, 37
Time-of-flight MR angiography, technique for,
115 116
Time-resolved two- and three-dimensional digital
subtraction angiography, technique for, 139
Pulse sequences, in MR angiography, 32 33,

121 123
in MR imaging. See Magnetic resonance imaging.
Tubo-ovarian complex, MR imaging of, 185 186
Tumors, bladder, MR imaging of. See Bladder.
Renal artery stenosis, MR angiography of, 130
Unicornuate uterus, MR imaging of, 181
Renal cell carcinoma, MR imaging of, 146 148
Urography, MR, technique for, 146
Renal hamartomas, MR imaging of, 148 150
Uterus didelphys, MR imaging of, 181
Renal transplant, postoperative evaluation of, MR

angiography in, 130 131, 133
Respiratory ordered phase encoding, in MR imaging,

45
Rest imaging, of cardiovascular disease, 17
Valvular heart disease, MR imaging of, 21

Ventricular function, MR imaging of, 25
Vesical congestion, of bladder, MR imaging of, 172

Radiologic Clinics of North America Body MR Imaging, Volume 41 Number 1 2003

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Radiologic Clinics of North America Body MR Imaging, Volume 41 Number 1 2003

Uploaded by

Copyright:

Available Formats

Radiol Clin N Am 41 (2003) xi

David A. Bluemke, MD, PhD

Body MR imaging applications continue to grow at

The complexity of these newer applications

Radiol Clin N Am 41 (2003) 1 15

MR physics of body MR imaging

Requirements and challenges of body imaging

Body MR imaging has progressed significantly in

Because many of the problems encountered in

E-mail address: todd.constable@yale.edu

R.T. Constable / Radiol Clin N Am 41 (2003) 115

mines the FOV and, in part, the noise present in the

R.T. Constable / Radiol Clin N Am 41 (2003) 115

somewhere in between in the oscillatory cycle at a

phase errors that arise due to motion. The next

Conventional image acquisition

Motion artifacts: motion

R.T. Constable / Radiol Clin N Am 41 (2003) 115

k-space data is collected starting at the bottom of the

R.T. Constable / Radiol Clin N Am 41 (2003) 115

tions from the pattern initially sampled will lead to

Imaging faster: pulse sequences

R.T. Constable / Radiol Clin N Am 41 (2003) 115

that a single excitation pulse and a single TR

these sequences are discussed in detail in the sections

R.T. Constable / Radiol Clin N Am 41 (2003) 115

ducing blurring and reducing imaging time. A second

effective TE, the primary signal change will be in

Fig. 5. Hyperecho pulse sequence diagram. By decreasing

R.T. Constable / Radiol Clin N Am 41 (2003) 115

phase-encode gradient is different for each echo,

amplitude that would be obtained if all pulses in the

R.T. Constable / Radiol Clin N Am 41 (2003) 115

bandwidth used for data acquisition with these

arising from breathing are shown in Fig. 8A, and

Fig. 7. Ghosts from phase errors between shots in multishot

R.T. Constable / Radiol Clin N Am 41 (2003) 115

with each shot, thereby constantly updating the

the gradient hardware capabilities to the maximum to

Imaging faster: collect less data

R.T. Constable / Radiol Clin N Am 41 (2003) 115

allows for a reduction in the number of phase-encode

small, and FOV wrap around artifacts (also known

R.T. Constable / Radiol Clin N Am 41 (2003) 115

and simultaneous acquisition of spatial harmonics

R.T. Constable / Radiol Clin N Am 41 (2003) 115

receivers provide excellent SNR and the ability to

[3] Hennig J. K-space sampling strategies. Eur Radiol

R.T. Constable / Radiol Clin N Am 41 (2003) 115

[9] Xiang QS, Henkelman RM. Motion artifact reduction

Third Scientific Meeting of the Society of Magnetic

R.T. Constable / Radiol Clin N Am 41 (2003) 115

sampled echo data. J Magn Reson Imaging 1991;

Radiol Clin N Am 41 (2003) 17 28

Cardiac MR imaging is an effective method for

Ischemic heart disease

E. Castillo, D.A. Bluemke / Radiol Clin N Am 41 (2003) 1728

and therefore is used infrequently in routine clinical practice.

E. Castillo, D.A. Bluemke / Radiol Clin N Am 41 (2003) 1728

The second enhancement pattern is observed at

Coronary artery imaging

E. Castillo, D.A. Bluemke / Radiol Clin N Am 41 (2003) 1728

in-plane spatial resolution (  0.7 1 mm) has been

in-plane spatial resolution ( 0.7 1 mm) has been