Merkle and A b d o m i n a l I m a g i n g • Pe r s p e c t i v e

Dale
Abdominal
MRI at 3.0 T
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

A C E N T U
R Y O F

Abdominal MRI at 3.0 T:

MEDICAL IMAGING
The Basics Revisited
Elmar M. Merkle1 OBJECTIVE. The purpose of our article is to describe the underlying physics concepts of
Brian M. Dale2 abdominal MRI at 3.0 T and their impact on signal-to-noise ratio, susceptibility artifacts, chem-
ical shift artifacts, and dielectric effects.
Merkle EM, Dale BM CONCLUSION. Abdominal MR sequence protocols optimized for 1.5-T scanners
should not be transferred to 3.0 T without substantial modification. In addition, specific patient
groups—for example, large patients with ascites—are not well suited to undergo an abdominal
MRI study at 3.0 T.

ver the past 2 years, ultrahigh- other regions of the body. It is also not clear

O field-strength whole-body 3.0-T

Keywords: abdominal imaging, field strength, MRI, MRI
technique, physics
DOI:10.2214/AJR.05.0932
Received June 1, 2005; accepted after revision
July 20, 2005.
1Department of Radiology, Duke University Medical Center,
Duke North, Rm. 1417, Box 3808, Erwin Rd., Durham, NC
27710. Address correspondence to E. M. Merkle
(elmar.merkle@duke.edu).
2Siemens
Medical Solutions, USA, Cary, NC 27519.
CME
This article is available for 1 CME credit. See supplemental
data for this article at www.ajronline.org or visit
www.arrs.org for more information.
AJR 2006; 186:1524–1532
0361–803X/06/1866–1524
© American Roentgen Ray Society
MR systems have been installed
in numerous institutions and are
being increasingly used in clinics. Besides
market considerations—for example, stra-
tegic investment to make a market state-
ment and to stay competitive—the main
reason to purchase an ultrahigh-field MR
system is the anticipated twofold MR sig-
nal-to-noise ratio (SNR) compared with a
standard 1.5-T MR scanner. This gain in
SNR can be kept or traded for either speed,
spatial resolution, or both. Although the
number of accessory receiving coils has
been limited in the past, the spectrum of
dedicated receiver coils offered by vendors
has increased significantly over the past 18
months, which allows almost all standard
MRI examinations to be performed on a
3.0-T whole-body MR system. Although ul-
trahigh-field MR systems have already been
shown to be advantageous for various indi-
cations in the brain and musculoskeletal
system compared with standard high-field
1.5-T MR systems, only a few scientific
studies have been published describing the
use of 3.0-T MR systems in the chest, abdo-
men, and pelvis [1–5]. Unfortunately, in-
sights gained in musculoskeletal or neu-
roimaging research at 3.0 T cannot simply
be transferred to body MRI because MR se-
quence protocols and object sizes differ sig-
nificantly in abdominal imaging. In addi-
tion, some artifacts are unique to ultrahigh-
field abdominal MRI and are not seen in
which patient groups will benefit from an
ultrahigh-field abdominal MRI study and
which patient groups should remain on a
1.5-T MR scanner. This article will illus-
trate the underlying physics concepts of ab-
dominal MRI at 3.0 T and their impact on
SNR, susceptibility artifacts, chemical shift
artifacts, and dielectric effects. On the basis
of these fundamental considerations, basic
recommendations will be provided for
which patient groups will likely benefit
from an ultrahigh-field MRI study and
which patient groups should undergo a stan-
dard 1.5-T abdominal MRI examination.
3.0 T Offers Twice the SNR:
A Persistent Myth
The idea that twice the magnetic field will
give twice the SNR is appealing, and at first it
seems correct because the intrinsic SNR in
MRI is approximately proportional to the
main magnetic field strength, B0 (equations 1
and 2) [6]. The equation for spin-echo–based
MRI sequences is
N PE N PA NAV
- ( 1 – e –TR ⁄ T1 )e –TE ⁄ T2
SNRSE ∝ B0 V ------------------------------- (1)
BW
where SNRSE = signal-to-noise ratio for a spin-
echo pulse sequence, B0 = main magnetic field
strength, V = voxel volume, NPE = number of
acquired phase encode lines, NPA = number of
acquired partitions, NAV = number of signals
averaged, BW = receiver bandwidth per pixel,
T1 = longitudinal relaxation time, and T2 =
transverse relaxation time.

1524 AJR:186, June 2006

 Abdominal MRI at 3.0 T

The equation for gradient-echo–based MRI which would further reduce the gain in SNR at msec for kidney, 493 msec for liver), but at 3.0
sequences is ultrahigh-field-strength MRI (Fig. 1B) for T that difference shrinks to 21% (774 msec
long TE protocols. Given the optimistic as- for kidney, 641 msec for liver). For other tis-
N PE N PA N AV sin ( θ ) ( 1 – e –TR ⁄ T1 ) *
- -------------------------------------------------- e –TE ⁄ T2 (2)
SNRGRE ∝ B0 V ------------------------------- sumption that the transverse relaxation time sue pairs, the relative T1 dispersion may actu-
BW ( 1 – e – TR ⁄ T1 cos ( θ ) )
(T2) is independent of the main magnetic field ally increase at ultrahigh field strength, rather
where SNRGRE = signal-to-noise ratio for a strength and assuming only an increase of the than decreasing as shown here for kidney and
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

spoiled gradient-echo sequence and θ = flip longitudinal relaxation time (T1), equations 1 liver. In any case, this example should illus-
angle. and 2 can be used to determine the theoretic trate why the contrast between various tissues
Note that, in both equations 1 and 2, the maximum relative gain in SNR during MRI of on T1-weighted images at 3.0-T MRI cannot
term under the square root is simply the total the liver. For turbo spin-echo–based T2- be identical to the contrast seen on standard
time spent acquiring data. Therefore, SNR is weighted sequences with sequential acquisi- 1.5-T T1-weighted MRI. Fortunately, on the
proportional to the main magnetic field tion such as HASTE sequences, an increase by basis of our current experience with more
strength, the voxel volume, the square root of a factor of approximately 1.8 in SNR can than 300 abdominal cases examined at ultra-
the total sampling time, and some sequence- be obtained. For gradient-echo–based T1- high-field MRI, this has not been a relevant
specific contrast-related terms. Some of these weighted sequences such as in- and opposed- clinical problem.
factors, such as the longitudinal relaxation phase and VIBE (volume interpolated breath- A second major factor with a negative im-
time (T1), receiver bandwidth, and specific hold examination) sequences, an increase by a pact on the gain in SNR is related to the spe-
absorption rate limitations, can affect the factor of approximately 1.6–1.7 in SNR can be cific absorption rate (SAR). When the main
SNR in a somewhat complicated manner by obtained. Thus, the theoretic twofold increase magnetic field strength is doubled, the SAR,
impacting other sequence-specific parame- in SNR at 3.0 T compared with 1.5 T will not a measure for energy deposition within the
ters (e.g., TR, flip angle). generally be obtained without sequence pa- human body, increases by a factor of 4. Al-
The longitudinal relaxation time, T1, in- rameter optimization. though the energy deposited at 3.0 T is still
creases at a higher magnetic field strength, In addition to the absolute change in the T1 nonionizing, the increased SAR requires an
which causes a decrease in SNR [1, 7] (see relaxation time as a function of magnetic field increased concern for patient safety. Because
equations 1 and 2) (Fig. 1A). The transverse strength, there are also relative changes for body MRI at 3.0 T almost always runs at the
relaxation time, T2, on the other hand, seems to which the T1 relaxation time for one tissue in- upper limits of the allowed SAR deposition,
be fairly independent of the main magnetic creases at a different rate from the T1 relax- patients are more likely to experience an un-
field strength [7]. However, one recently pub- ation time of another tissue. For example, ac- comfortable sensation of warmth or heating.
lished study by de Bazelaire et al. [1] suggests cording to Bottomley et al. [7], at 1.5 T the T1 In addition, protocol adjustments are fre-
a marked decrease of the transverse relaxation relaxation time of the kidney is 32% greater quently necessary, such as an increase of the
time (T2) at higher magnetic field strengths, than the T1 relaxation time of the liver (652 TR, a decrease in the number of slices, or a

800
80

600
T2 (msec)
T1 (msec)

60

400
40

200

20
0
0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1.0 1.5 2.0 2.5 3.0
Field Strength (T) Field Strength (T)
A B
Fig. 1—Diagrammatic illustrations of relationships of relaxation times of liver tissue and main magnetic field strength (B0). Note that data of de Bazelaire et al. [1] are within
confidence interval obtained by Bottomley et al. [7] even though 3.0-T data point is outside scope of article by Bottomley et al.
A, Graph shows relationship of longitudinal relaxation time, T1, of liver tissue and main magnetic field strength. Data are shown from meta-analysis based on more than 800
study samples performed by Bottomley et al. [7] (■), together with data based on six volunteers acquired by de Bazelaire et al. [1] (▲). Note nonlinear increase of T1 of liver
tissue with main magnetic field strength. Also shown are nonlinear regression (solid line) and 95% confidence interval for data (dashed lines) as described by Bottomley et al.
B, Graph shows relationship of transverse relaxation time, T2, of liver tissue and main magnetic field strength. Data are shown from meta-analysis based on more than 250
study samples performed by Bottomley et al. [7] (▲), together with data based on six volunteers acquired by de Bazelaire et al. [1] (■). Note that no obvious relationship is
seen between T2 of liver tissue and main magnetic field strength. Also shown are mean (solid line) and 95% confidence interval (dashed lines) for data as described by
Bottomley et al.

AJR:186, June 2006 1525

 Merkle and Dale
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

A B

C D
Fig. 2—Chemical shift artifacts of the first and second kinds at various magnetic field strengths in same patient, 32-year-old woman.
A, Axial gradient-echo in-phase image acquired at field strength of 1.5 T shows minimal chemical shift artifacts of the first kind along frequency-encoding axis.
B, Axial gradient-echo in-phase image acquired at field strength of 3.0 T shows marked chemical shift artifacts of the first kind along frequency-encoding axis that appear
bright toward higher part of readout gradient field (long arrow) and dark along lower part (short arrow).
C, Axial gradient-echo opposed-phase image acquired at field strength of 1.5 T shows chemical shift artifact of the second kind in all pixels along fat–water interface.
D, Axial gradient-echo opposed-phase image acquired at field strength of 3.0 T shows chemical shift artifact of the second kind, which is not significantly different from
artifact seen at standard magnetic field strength of 1.5 T (compare with C).

decrease of the flip angle. These adjustments
are all undesirable because they increase
scanning time, reduce anatomic coverage, al-
ter contrast, or further reduce the gain in SNR
at 3.0 T when compared with a standard 1.5-
T MRI system. Finally, much of the radiofre-
quency transmitter and receiver technology at
1.5 T is relatively mature compared with the
newer technology at 3.0 T.
All these reasons contribute to a gain in
SNR that is less than the factor of 2.0 origi-
nally expected. This may help explain why
Morakkabati-Spitz et al. [4] found that, “Vi-
sual signal to noise was rated equal for all 19
pelvic MRI studies” when applying a proto-
col with similar spatial and temporal resolu-
tion at both field strengths.
Chemical Shift Artifacts at 3.0 T:
A Double-Edged Sword
The chemical shift artifact of the first kind
is due to a difference in the resonant fre-
quency between water and fat and is seen only
along the frequency-encoding axis and the
slice-selection dimension [8]. This difference
in resonant frequency is directly proportional
to the main magnetic field strength and has
been measured as approximately 3.5 ppm, re-
sulting in a difference of about 225 Hz at 1.5
T, or a difference of about 450 Hz at 3.0 T.
This difference causes a chemical shift mis-
registration, which is most easily seen around
the kidneys (Fig. 2). The chemical shift arti-
fact of the first kind appears as a hypointense
band, 1 to several pixels in width, toward the
lower part of the readout gradient field, and as
a hyperintense band toward the higher part of
the readout gradient field. At a constant field
of view, base resolution, and receiver band-
width, the chemical shift artifact of the first
kind will be twice as wide at 3.0 T as at stan-
dard 1.5-T imaging. Usually, this enlarged ar-

1526 AJR:186, June 2006

 Abdominal MRI at 3.0 T
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

A B
Fig. 3—Effect of receiver bandwidth on size of chemical shift artifacts of the first kind
at 3.0 T in 40-year-old man.
A, Axial in-phase T1-weighted gradient-echo image through kidney shows
significant chemical shift artifact (arrow) (receiver bandwidth, 210 Hz; pixel shift, 1.9;
SNR [signal-to-noise ratio]liver, 75).
B, Axial in-phase T1-weighted gradient-echo image through kidney with twice
receiver bandwidth as in A shows smaller chemical shift artifact (arrow) (receiver
bandwidth, 415 Hz; pixel shift, 1.0; SNRliver, 56). However, SNR is also decreased by
approximately 30%.
C, Axial in-phase T1-weighted gradient-echo image through kidney with four times
receiver bandwidth as in A shows markedly smaller chemical shift artifact (arrow)
(receiver bandwidth, 850 Hz; pixel shift, 0.5; SNRliver, 41). However, SNR is again
substantially decreased by another 30%.

tifact does not cause substantial problems in
clinical body MRI at 3.0 T. However, it may
be problematic in selected cases such as the
search for a subcapsular renal hematoma or
an intramural aortic hematoma. In these
cases, the receiver bandwidth can be in-
creased to minimize the chemical shift arti-
fact of the first kind. Unfortunately, this
comes at the expense of SNR: doubling the
receiver bandwidth will decrease the SNR by
approximately 30% (see equations 1 and
2: 1--2- ≈ 0.7 → –30%) (Fig. 3). Another option
is to repeat the MR pulse sequence with either
chemical shift fat saturation, inversion null-
ing, or water excitation, which will eliminate
chemical shift artifacts effectively, allow im-
aging at the lower bandwidth, and return the
30% loss in SNR.
The chemical shift artifact of the second
kind is not limited to the frequency-encod-
ing axis but may be seen in all pixels along
a fat–water interface because it is based on
an intravoxel phase-cancellation effect in
which fat and water exist in the same voxel
[8]. The size of this artifact does not in-
crease with the main magnetic field strength
and is defined by the spatial resolution of
the MR sequence (Fig. 2). However, the TE
needs to be adjusted because the frequency
difference is twice as large as with the stan-
dard 1.5-T MR systems, as described in the
section on chemical shift artifacts of the first
kind. Using a 3.0-T MR system, both fat and
water protons are in phase at 2.2, 4.4, 6.6
msec, and so on, and out of phase (also re-
ferred to as opposed phase) at 1.1, 3.3, 5.5
msec, and so on. Note that at 1.5 T, the fat
and water are phase-opposed at 2.2 msec
and in phase at 4.4 sec (nominal values). In
short, by doubling the field strength we have
halved the TEs for in-phase and opposed-
phase imaging.
Fortunately, the increased difference in res-
onant frequency between water and fat at 3.0
T may also be advantageous because it allows
a better separation of the fat and water peak
during MR spectroscopy, and a better or faster
fat suppression using other chemical shift
techniques as well—for example, fat satura-
tion and water excitation.
Susceptibility Artifacts: A Closer Look
Magnetic susceptibility is the extent to
which a material becomes magnetized when

AJR:186, June 2006 1527

 Merkle and Dale
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

Fig. 4—Effect of magnetic field strength on size of susceptibility artifacts in various MR sequences in vitro.
A, Photograph shows phantom setup with water-filled straw (arrow) embedded in gelatin. Three pairs of surgical clips (arrowheads) are embedded at various distances from straw.
B, HASTE image acquired at 1.5 T (TR/TE, 1,010/128; field of view, 250 mm2; slice thickness, 2.8 mm; matrix, 256 × 205; bandwidth, 490 Hz) shows typical susceptibility artifacts
caused by surgical clips.
C, HASTE image acquired at 3.0 T (1,010/128; field of view, 250 mm2; slice thickness, 2.8 mm; matrix, 256 × 205; bandwidth, 490 Hz) shows larger susceptibility artifacts caused
by surgical clips when compared with B.
D, Gradient-echo image acquired at 1.5 T (118/2.4; field of view, 280 × 210 mm2; slice thickness, 3.0 mm; matrix, 256 × 154; bandwidth, 385 Hz) shows typical susceptibility
artifacts caused by surgical clips.
E, Gradient-echo image acquired at 3.0 T (118/2.4; field of view, 280 × 210 mm2; slice thickness, 3.0 mm; matrix, 256 × 154; bandwidth, 385 Hz) shows larger susceptibility
artifacts caused by surgical clips when compared with D. Note that artifact size increases by approximately 100% in terms of volume.

A B
Fig. 5—Effect of magnetic field strength on size of metal-related susceptibility artifacts in vivo in 58-year-old man.
A, Gradient-echo opposed-phase image acquired at 1.5 T (TR/TE, 200/2.2) shows typical susceptibility artifacts caused by surgical clips (arrows).
B, Gradient-echo opposed-phase image acquired at 3.0 T (200/1.5) shows larger susceptibility artifacts (arrows) caused by surgical clips when compared with A despite
shorter TE.

1528 AJR:186, June 2006

 Abdominal MRI at 3.0 T
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

A B
Fig. 6—Negative effect of magnetic field strength on size of gas-related susceptibility artifacts in vivo in 52-year-old man.
A, Gradient-echo in-phase image acquired at 1.5 T (TR/TE, 200/4.4) shows minor susceptibility artifacts (arrows) in hepatic flexure and transverse colon caused by colonic gas.
B, Gradient-echo in-phase image acquired at 3.0 T (200/4.4) shows larger susceptibility artifacts (arrows) in colon when compared with A. Note that these susceptibility
artifacts obscure colonic wall.

A B
Fig. 7—Positive effect of magnetic field strength on size of gas-related susceptibility artifacts in vivo in 65-year-old woman.
A, Gradient-echo in-phase image acquired at 1.5 T (TR/TE, 200/4.4) shows minor susceptibility artifact (arrow) in left hepatic bile duct caused by pneumobilia that may be
misinterpreted as a branch of portal venous system.
B, Gradient-echo in-phase image acquired at 3.0 T (200/4.4) shows markedly larger susceptibility artifact (arrow) in left hepatic bile duct caused by pneumobilia when
compared with A. Enlarged susceptibility artifact makes misinterpretation less likely.

placed in a magnetic field. Susceptibility ar-
tifacts occur as the result of microscopic gra-
dients or variations in the magnetic field
strength that occur near the interfaces of ma-
terials of different magnetic susceptibility.
These artifacts are usually caused by metal-
lic objects from previous surgical or inter-
ventional procedures near or in the imaging
field of view because the susceptibility of
metal is much higher than that of soft tissue.
Susceptibility artifacts increase with the
main magnetic field strength and are approx-
imately twice as large in terms of volume at
3.0 T as at standard 1.5-T MRI [9] (Fig. 4).
This may be advantageous in selected cases
because metal-related susceptibility artifacts
from surgical clips or surgical debris—for
example, prior cholecystectomy or prior he-
patic resection—may be better seen (Fig. 5).
However, it is possible that enlarged suscep-
tibility artifacts may obscure important find-

AJR:186, June 2006 1529

 Merkle and Dale

Fig. 8—Physical basis of radiofrequency shielding. In step A, rapidly varying

magnetic field (black arrows) induces a circulating electric field (white arrows). In
presence of a conductive medium (step B), circulating electric field leads to a
circulating current (gray arrows). In step C, circulating current acts as
electromagnet to produce magnetic field in opposite direction; and in step D,
amplitude of overall magnetic field is reduced. Note that steps B and C require a
conductive medium and that effect is stronger in more conductive medium. In
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved

imaging, this effect can be noticed in patients with ascites or fetal imaging, in which
circulating currents can be established in relatively large regions of highly
conductive fluid. Resulting artifacts are generally more visible in sequences that use
a large number of radiofrequency pulses to generate contrast.

Fig. 9—Severe standing wave and conductivity artifact in 38-year-old woman with
liver cirrhosis and ascites during ultrahigh-field-strength MRI at 3.0 T.
A, Coronal HASTE image shows marked signal loss in center of image (arrows). Fluid
accumulations in peritoneal cavity enlarge abdomen and increase electrical
conductivity in field of view, causing severe artifacts.
B, Axial HASTE image again shows marked signal loss in center of image (arrows)
and represents severe standing wave and conductivity artifact.
C, Contrast-enhanced axial gradient-echo T1-weighted image acquired at same
level as B shows normal anatomy in center of field of view (arrows) and no evidence
of susceptibility artifacts.

A C

ings at 3.0-T MRI that may have been visu- vices that are considered MR safe at a field rigorously tested at 3.0 T as well before af-
alized at standard 1.5-T MRI. It must be strength of 1.5 T are not necessarily safe at fected patients can undergo an MRI exami-
clearly stated here that metal-containing de- 3.0 T [10–15]. All these devices need to be nation at this field strength.

1530 AJR:186, June 2006


Fig. 10—Severe standing
wave and conductivity
artifact in 28-year-old
pregnant woman during
fetal ultrahigh-field-
strength MRI at 3.0 T.
Coronal HASTE image
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved
shows marked signal
loss (long arrows). This
artifact is caused by
large amount of amniotic
fluid and increased size
of abdomen, which
increase electrical
conductivity in field of
view. Short thin arrows
mark placenta; thick
arrows mark fetal torso.
(Courtesy of Jones B,
Cincinnati, OH)
Susceptibility artifacts also occur next to
gas-filled structures, such as the gas-filled
bowel, because the susceptibility of gas is
much less than that of soft tissue (Fig. 6).
Thus, bowel wall imaging in patients with in-
flammatory bowel disease or patients referred
for MR colonography seems to be more chal-
lenging at 3.0 T. However, enlarged suscepti-
bility artifacts due to a gas–soft-tissue inter-
face may also be helpful in detecting gas—for
example, intrahepatic pneumobilia or free in-
traperitoneal gas (Fig. 7).
Standing Wave Effects
In addition to the exacerbation of artifacts
that are seen at 1.5 T, some new artifacts also
begin to appear at 3.0 T. These artifacts are re-
lated to the higher frequency B1 transmit fields
that are used at 3.0 T. The wavelength of the ra-
diofrequency field at 128 MHz is 234 cm in free
space, which is much larger than the field of
view for clinical body imaging. However, water
(and most body tissue) has a rather high dielec-
tric constant, which reduces both the speed and
wavelength of electromagnetic radiation. For
visible light, this effect causes a straight stick
entering water at an angle to appear bent. For
MRI, this effect reduces the radiofrequency
field wavelength from 234 cm in free space to
about 30 cm in most human tissues—that is,
water-containing tissues [16]. This size is ap-
proximately the size of the field of view for
many body applications and can result in a so-
called standing wave effect (often incorrectly
called a “dielectric resonance” effect) [17]. As
AJR:186, June 2006
Abdominal MRI at 3.0 T
a result, strong signal variations across an im-
age can be seen, especially brightening or dark
“holes” in regions away from the receive coil
caused by constructive or destructive interfer-
ence from the standing waves. These artifacts
become more pronounced the larger the region
of interest is relative to the wavelength—that is,
they are seen more in obese patients with a dis-
tended abdomen than in thin patients.
A rapidly changing magnetic field, like the
radiofrequency transmit field, will induce a
circulating electric field (Fig. 8). When this
happens in a conductive medium, a circulat-
ing electric current is established. This cur-
rent in turn acts like an electromagnet that op-
poses the changing magnetic field, reducing
the amplitude and dissipating the energy of
the radiofrequency field. The more conduc-
tive the medium, the stronger the opposing
electromagnet and therefore the greater the
attenuation of the radiofrequency field. In
construction of the MRI suite, this principle is
used by encasing the room in a copper con-
ductor. Because of the high conductivity of
copper, any incoming radiofrequency waves
are almost completely attenuated and the
magnet is shielded from external interference.
To a lesser extent, large amounts of relatively
highly conductive tissues can cause similar
shielding effects, resulting in hypointense ar-
eas in the image where the radiofrequency
field is partially attenuated [16].
These two effects combine to cause partic-
ularly strong artifacts for 3.0-T body MRI in
pregnant patients and in patients with ascites
(Figs. 9 and 10). In both cases, not only are
the standing wave effects more pronounced
because of the enlarged abdomen, but greater
radiofrequency field attenuation is also
present because of the increased amounts of
highly conductive amniotic or ascitic fluid.
Summary and General
Recommendations
Body MRI at 3.0 T is still in its infancy and
will improve substantially over the next sev-
eral years. However, radiologists need to
know these several limitations based on the
laws of physics:
First, overall, the gain in SNR at 3.0 T will
be less than twofold compared with a stan-
dard 1.5-T MR system because of the ines-
capable increase of the longitudinal relax-
ation time T1. Also, the increased SAR
deposition at ultrahigh field strength often
requires protocol adjustments that can fur-
ther reduce the anticipated gain. The gain in
SNR will be higher in T2-weighted se-
quences than in T1-weighted sequences be-
cause longer TRs allow a more complete re-
covery of the longitudinal magnetization,
and transverse relaxation times (T2) are
fairly independent of the main magnetic
field strength. Thus, patients referred for MR
cholangiography may benefit from an ultra-
high-field-strength MR examination.
Second, chemical shift artifacts of the first
kind are twice as large on ultrahigh-field MRI
as on standard 1.5-T MRI. Chemical shift arti-
facts of the second kind, on the other hand, do
not increase in size, although the timing is al-
tered. Fortunately, the increased difference in
resonant frequency between water and fat at
3.0 T is also advantageous because it allows a
better separation of the fat and water peaks dur-
ing MR spectroscopy and a better or faster fat
suppression using chemical shift techniques.
Third, susceptibility artifacts are twice as
big on 3.0-T MRI. Although patients referred
for a colon study may be challenging, the
search for gas—for example, free air or pneu-
mobilia—should be easier on 3.0-T MRI. Pa-
tients with metal implants should undergo an
MR examination at 3.0 T only if the metal-
containing device has been proven to be MR
safe for this field strength.
Fourth, standing wave and conductivity ef-
fects are usually not seen at a field strength of
1.5 T. At 3.0 T, these artifacts are most pro-
nounced in pregnant women in the second and
third trimesters because of the large amount of
amniotic fluid and the increased size of the ab-
domen. Fetal MRI should therefore not be per-
1531

formed at 3.0 T because of these severe arti-
facts and the increased safety concerns. The
same holds true for patients with a large
amount of ascites, who are also not well suited
for an ultrahigh-field MRI examination.
Finally, most patients can undergo an ab-
Downloaded from www.ajronline.org by 202.80.217.171 on 11/22/16 from IP address 202.80.217.171. Copyright ARRS. For personal use only; all rights reserved
dominal MRI study at 3.0 T with a reasonable
outcome in terms of image quality.
Acknowledgments
We thank David E. Purdy and H. Cecil
Charles for reviewing the manuscript and the
subsequent stimulating discussions.
References
1. de Bazelaire CM, Duhamel GD, Rofsky NM, Alsop
DC. MR imaging relaxation times of abdominal and
pelvic tissues measured in vivo at 3.0 T: preliminary
results. Radiology 2004; 230:652–659
2. Sosna J, Rofsky NM, Gaston SM, DeWolf WC,
Lenkinski RE. Determinations of prostate volume
at 3-Tesla using an external phased array coil: com-
parison to pathologic specimens. Acad Radiol
2003; 10:846–853
3. Katz-Brull R, Rofsky NM, Lenkinski RE. Breath-
hold abdominal and thoracic proton MR spectros-
copy at 3T. Magn Reson Med 2003; 50:461–467
4. Morakkabati-Spitz N, Gieseke J, Kuhl C, et al. 3.0-
T high-field magnetic resonance imaging of the fe-
F O R YO U R I N F O R M AT I O N
This article is available for CME credit. See supplemental data for this
article at www.ajronline.org or visit www.arrs.org for more information.
1532
Merkle and Dale
male pelvis: preliminary experiences. Eur Radiol
2005; 15:639–644
5. Greenman RL, Shirosky JE, Mulkern RV, Rofsky
NM. Double inversion black-blood fast spin-echo
imaging of the human heart: a comparison between
1.5 T and 3.0 T. J Magn Reson Imaging 2003;
17:648–655
6. Edelstein WA, Glover GH, Hardy CJ, Redington
RW. The intrinsic signal-to-noise ratio in NMR im-
aging. Magn Reson Med 1986; 3:604–618
7. Bottomley PA, Foster TH, Argersinger RE, Pfeifer
LM. A review of normal tissue hydrogen NMR re-
laxation times and relaxation mechanisms from
1–100 MHz: dependence on tissue type, NMR fre-
quency, temperature, species, excision, and age.
Med Phys 1984; 11:425–448
8. Elster AE, Burdette JH. Questions and answers in
magnetic resonance imaging, 2nd ed. St. Louis,
MO: Mosby, 2001:6, 128
9. Lewin JS, Duerk JL, Jain VR, Petersilge CA, Chao
CP, Haaga JR. Needle localization in MR-guided
biopsy and aspiration: effects of field strength, se-
quence design, and magnetic field orientation. AJR
1996; 166:1337–1345
10. Sommer T, Maintz D, Schmiedel A, et al. High field
MR imaging: magnetic field interactions of aneu-
rysm clips, coronary artery stents and iliac artery
stents with a 3.0 Tesla MR system [in German].
Rofo 2004; 176:731–738
11. Shellock FG. Biomedical implants and devices: as-
sessment of magnetic field interactions with a 3.0-
Tesla MR system. J Magn Reson Imaging 2002;
16:721–732
12. Shellock FG, Tkach JA, Ruggieri PM, Masaryk TJ.
Cardiac pacemakers, ICDs, and loop recorder: eval-
uation of translational attraction using conventional
(“long-bore”) and “short-bore” 1.5- and 3.0-Tesla
MR systems. J Cardiovasc Magn Reson 2003;
5:387–397
13. Baker KB, Nyenhuis JA, Hrdlicka G, Rezai AR,
Tkach JA, Shellock FG. Neurostimulation systems:
assessment of magnetic field interactions associ-
ated with 1.5- and 3-Tesla MR systems. J Magn Re-
son Imaging 2005; 21:72–77
14. Shellock FG, Gounis M, Wakhloo A. Detachable
coil for cerebral aneurysms: in vitro evaluation of
magnetic field interactions, heating, and artifacts at
3T. AJNR 2005; 26:363–366
15. Shellock FG, Forder JR. Drug eluting coronary
stent: in vitro evaluation of magnet resonance safety
at 3 Tesla. J Cardiovasc Magn Reson 2005;
7:415–419
16. Haacke EM, Brown RW, Thompson MR, Venkate-
san R. Magnetic resonance imaging: physical prin-
ciples and sequence design. New York, NY: Wiley,
1999:10, 662
17. Collins CM, Liu W, Schreiber W, Yang QX, Smith
MB. Central brightening due to constructive inter-
ference with, without, and despite dielectric reso-
nance. J Magn Reson Imaging 2005; 21:192–196
AJR:186, June 2006