See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/260603026

Biomagnetism

Chapter · January 2000

CITATIONS READS

0 43

1 author:

Jaakko Malmivuo
Technische Universität Berlin
637 PUBLICATIONS 3,422 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Internet education View project

11th International Congress of Bioelectromagnetism 23.-26.5.2018, Aachen, Germany View project

All content following this page was uploaded by Jaakko Malmivuo on 18 September 2017.

The user has requested enhancement of the downloaded file.

Malmivuo, J. “Biomagnetism.”
The Biomedical Engineering Handbook: Second Edition.
Ed. Joseph D. Bronzino
Boca Raton: CRC Press LLC, 2000
 16
Biomagnetism

16.1 Theoretical Background

Origin of Bioelectric and Biomagnetic Signals •
Measurement of the Biomagnetic Signals • Independence
of Bioelectric and Biomagnetic Signals
16.2 Sensitivity Distribution of Dipolar Electric and
Magnetic Leads
Concepts of Lead Vector and Lead Field • Lead Fields of
Leads Detecting the Electric and Magnetic Dipolar Moments
of a Volume Source • Independence of Dipolar Electric and
Magnetic Leads
16.3 Magnetocardiography (MCG)
Selection of the Source Model for MCG • Detection of the
Equivalent Magnetic Dipole of the Heart • Diagnostic
Performance of ECG and MCG • Technical Reasons to Use
the MCG • Theoretical Reasons to Use the MCG
16.4 Magnetoencephalography (MEG)
Sensitivity Distribution of the Axial Magnetometer •
Sensitivity Distribution of the Planar Gradiometer • Half-
Jaakko Malmivuo Sensitivity Volumes of Electro- and Magnetoencephalography
Ragnar Granit Institute, Tampere Leads • Sensitivity of EEG and MEG to Radial and
University of Technology Tangential Sources

Since the first detection of the magnetocardiogram (MCG) in 1963 by Baule and McFee [Baule and
McFee, 1963], new diagnostic information from biomagnetic signals has been widely anticipated. The
first recording of the magnetoencephalogram (MEG) was made in 1968 by David Cohen [Cohen, 1968],
but it was not possible to record biomagnetic signals with good signal quality before the invention of
the superconducting quantum interference device (SQUID) in 1970 [Zimmerman et al., 1970].

16.1 Theoretical Background

Origin of Bioelectric and Biomagnetic Signals

In 1819, Hans Christian Örsted demonstrated that when an electric current flows in a conductor, it
generates a magnetic field around it (Örsted, 1820]. This fundamental connection between electricity
and magnetism was expressed in exact form by James Clerk Maxwell in 1864 [Maxwell, 1865]. In
bioelectromagnetism, this means that when electrically active tissue produces a bioelectric field, it simul-
taneously produces a biomagnetic field as well. Thus the origin of both the bioelectric and the biomagnetic
signals is the bioelectric activity of the tissue.

© 2000 by CRC Press LLC

 The following equations describe the electric potential field (16.1) and the magnetic field (16.2) of a
→
volume source distribution J i in an inhomogeneous volume conductor. The inhomogeneous volume
conductor is represented by a piecewise homogeneous conductor where the regions of different conduc-
tivity σ are separated by surfaces S.

   
( ) ∫ J ⋅∇  1r  dv + ∑ ∫ (σ′′− σ′ ) Φ∇  1r  dS
4 πσΦ r =
v
i

Sj
j j j (16.1)
j

 1  
( ) ∫J
4 πH r =
v
i
× ∇   dv +
 r
∑ ∫ (σ′′− σ′ ) Φ∇  1r  dS
Sj
j j j (16.2)
j

The first term on the right-hand side of Eqs. (16.1) and (16.2) describes the contribution of the volume
source, and the second term describes the contribution of boundaries separating regions of different
conductivity, i.e., the contribution of the inhomogeneities within the volume conductor. These equations
were developed by David Geselowitz [Geselowitz, 1967, 1970].

Measurement of the Biomagnetic Signals

The amplitude of the biomagnetic signals is very low. The strongest of them is the MCG, having an
amplitude on the order of 50 pT. This is roughly one-millionth of the static magnetic field of the earth.
The amplitude of the MEG is roughly 1% of that of the MCG. This means that, in practice, the MEG
can only be measured with the SQUID and that the measurements must be done in a magnetically
shielded room. The MCG, instead, can be measured in the clinical environment without magnetic
shielding.

Independence of Bioelectric and Biomagnetic Signals

The source of the biomagnetic signal is the electric activity of the tissue. Therefore, the most interesting
and most important question in biomagnetism is whether the biomagnetic signals contain new infor-
mation that cannot be obtained from bioelectric signals; in other words, whether the bioelectric and
biomagnetic signals are fully independent or whether there is some interdependence. If the signals were
fully independent, the biomagnetic measurement would possibly give about the same amount of new
information as the bioelectric method. If there were some interdependence, the amount of new infor-
mation would be reduced.
Helmholtz’s theorem states that “A general vector field, that vanishes at infinity, can be completely
represented as the sum of two independent vector fields, one that is irrotational (zero curl) and another
that is solenoidal (zero divergence)” [Morse and Feshbach, 1953; Plonsey and Collin, 1961]. The
—

impressed current density J i is a vector field that vanishes at infinity and, according to the theorem, may
be expressed as the sum of two components:

J i = J Fi + J Vi (16.3)

where the subscripts F and V denote flow and vortex, respectively. By definition, these vector fields satisfy
— —

 × J Fi = 0 and  · J Vi = 0. We first examine the independence of the electric and magnetic signals in the
infinite homogeneous case, when the second term on the right-hand side of Eqs. (16.1) and (16.2), caused
by inhomogeneities, is zero. The equation for the electric potential may be rewritten as

© 2000 by CRC Press LLC

  1 ∇⋅ J i
4 πσΦ =
∫v
∇   ⋅ J i dv =
 r ∫ v r
dv (16.4)

and that for the magnetic field may be rewritten as

 1 ∇× Ji
v ∫
4 πH = − ∇   × J i dv = −
 r ∫ v r
dv . (16.5)

Substituting Eq. (16.3) into Eqs. (16.4) and (16.5) shows that under homogeneous and unbounded
→
conditions, the bioelectric field arises from  · J Fi , which is the flow source, and the biomagnetic field
→
arises from  × J Vi , which is the vortex source. For this reason, in the early days of biomagnetic research
it was generally believed that the bioelectric and biomagnetic signals were fully independent. However,
it was soon recognized that this could not be the case. For example, when the heart beats, it produces
an electric field recorded as the P, QRS, and T waves of the ECG, and it simultaneously produces the
corresponding magnetic waves recorded as the MCG. Thus the ECG and MCG signals are not fully
independent.
There have been several attempts to explain the independence/interdependence of bioelectric and
biomagnetic signals. Usually these attempts discuss different detailed experiments and fail to give a
satisfying general explanation. This important issue may be easily explained by considering the sensitivity
distributions of the ECG and MCG lead systems, and this will be discussed in the next section.

16.2 Sensitivity Distribution of Dipolar Electric

and Magnetic Leads

Concepts of Lead Vector and Lead Field

Lead Vector
Let us assume that two electrodes (or sets of electrodes) are placed on a volume conductor to form a
lead. Let us further assume that inside the volume conductor in a certain location Q there is placed a
unit dipole consecutively in the x, y, and z directions (Fig. 16.1a). Due to the sources, we measure from
the lead the signals cx , cy , and cz , respectively. Due to linearity, if instead of the unit dipoles we place in
the source location dipoles that are px , py , and pz times the unit vectors, we measure signals that are cx px ,
cy py , and cz pz , respectively.
If these dipoles are placed simultaneously to the source location, due to the principle of superposition,
we measure from the lead a voltage that is

V = c x p x + c y p y + c z pz . (16.6)

– – – – –
These dipoles can be considered to be components of a dipole p, that is, p = px i + py j + pz k. We may
– – – – –
understand the coefficients cx , cy , and cz to be components of a vector c, that is, c = cx i + cy j + cz k. Now
– –
we may express the lead voltage Eq. (16.6) as the scalar product of the vector c and the dipole p as

V =c⋅p. (16.7)
– –
The vector c is a three-dimensional transfer coefficient that describes how a dipole source p at a fixed
point Q inside a volume conductor influences the voltage measured from the lead and is called the lead
vector.

© 2000 by CRC Press LLC

 FIGURE 16.1 The concepts of (a) lead vector and (b) lead field. See the text for more details.

–
The lead vector c describes what is the sensitivity of the lead to a source locating at the source location.
It is self-evident that for another source location the sensitivity may have another value. Thus the
sensitivity, i.e., the lead vector, varies as a function of the location, and we may say that it has a certain
distribution in the volume conductor. This is called the sensitivity distribution.
Lead Field
We may define the value of the lead vector at every point in the volume conductor. If we then place the
lead vectors to the points for which they are defined, we have –a field of lead vectors throughout the
volume conductor. This field of lead vectors is called the lead field JL. The lead field illustrates the behavior
of the sensitivity in the volume conductor and is a very powerful tool in analyzing the properties of
electric and magnetic leads (see Fig. 16.1b).
It follows from the principle of reciprocity, described by Hermann von Helmholtz in 1853 [Helmholtz,
1853], that the lead field is identical to the electric current field that arises in the volume conductor if a
unit current, called reciprocal current
–
IR, is fed to the lead.
When we –know the lead field JL, we can determine the signal VL in the lead due to the volume source
distribution J i . For each source element the signal is, of course, proportional to the dot product of the

© 2000 by CRC Press LLC

 FIGURE 16.1 (continued)

source element and the lead field at the source location, as shown in Eq. (16.7). The contributions of
the whole volume source is obtained by integrating this throughout the volume source. Thus the signal
the volume source generates to the lead is

∫ σ J ⋅J
1
VL = L
i
dv . (16.8)

The lead field may be illustrated either with lead vectors in certain locations in the volume conductor
or as the flow lines of the distribution of the reciprocal current in the volume conductor. This is called
the lead current field. In the latter presentation, the lead field current flow lines are oriented in the direction
of the sensitivity, and their density is proportional to the magnitude of the sensitivity.

© 2000 by CRC Press LLC

Lead Fields of Leads Detecting the Electric and Magnetic Dipole
Moments of a Volume Source
Electric Lead
The sensitivity of a lead system that detects the electric dipole moment of a volume source consists of
three orthogonal components (Fig. 16.2a). Each of these is linear and homogeneous. In other words,
one component of the electric dipole moment is detected when the corresponding component of all
–
elements of the impressed current density J i are detected with the same sensitivity throughout the source
area.
Magnetic Lead
The sensitivity distribution of a lead system that detects the magnetic dipole moment of a volume source
also consists of three orthogonal components (Fig. 16.2b). Each of these has such a form that the sensitivity
is always tangentially oriented around the symmetry axis (the coordinate axis). The magnitude of the
sensitivity is proportional to the radial distance from the symmetry axis and is zero on the symmetry axis.

Independence of Dipolar Electric and Magnetic Leads

Electric Lead
The sensitivity distributions of the three components of the lead system detecting the electric dipole
moment of a volume source are orthogonal. This means that none of them can be obtained as a linear
combination of the two other ones. (Note that any fourth measurement having a similar linear sensitivity
distribution would always be a linear combination of the three previous ones.) Thus the sensitivity
distributions, i.e., the leads, are orthogonal and thus independent. However, because the three electric
signals are only different aspects of the same volume source, they are not (fully) independent.
Magnetic Lead
The sensitivity distributions of the three components of the lead system detecting the magnetic dipole
moment of a volume source are also orthogonal, meaning that no one of them can be obtained as a
linear combination of the two other ones. Thus, similarly, as in measurement of the electric dipole
moment, the sensitivity distributions, i.e., the leads, are orthogonal and thus independent. However,
because the three magnetic signals are only different aspects of the same volume source, they are not
(fully) independent.
On the basis of the sensitivity distributions, we also can similarly explain the independence between
the electric and magnetic signals. According to Helmholtz’s theorem, the electric leads are orthogonal to
the three magnetic leads. This means that none of these six leads can be obtained as a linear combination
of the other five. However, the six signals, which they measure, are not independent because they arise
from the same electrically active volume source.

16.3 Magnetocardiography (MCG)

Selection of the Source Model for MCG

In ECG and MCG it is the clinical problem to solve the inverse problem, i.e., to solve the source of the
detected signal in order to get information about the anatomy and physiology of the source. Although
the actual clinical diagnostic procedure is based on measuring certain parameters, such as time intervals
and amplitudes, from the detected signal and actually not to display the components of the source, the
selection of the source model is very important from the point of view of available information.
In clinical ECG, the source model is a dipole. This is the model for both the 12-lead ECG and
vectorcardiography (VCG). In 12-lead ECG, the volume conductor (thorax) model is not considered,
which causes considerable distortion of the leads. In VCG, only the form of the volume conductor is

© 2000 by CRC Press LLC

FIGURE 16.2 Sensitivity distributions, i.e., lead fields of lead systems detecting (a) electric and (b) magnetic dipole
moments of a volume source. The lead field of the electric lead is shown both with vectors representing the magnitude
and direction of the lead field (on the left) and with lead field current flow lines (on the right).

© 2000 by CRC Press LLC

modeled. This decreases the distortion in the lead fields but does not eliminate it completely. Note that
today the display systems used in these ECG and VCG systems do not play any role in the diagnostic
procedure because the computerized diagnosis is always based on the signals, not on the display.
In selection of the source model for MCG, it is logical, at least initially, to select the magnetic source
model to be on the same theoretical level with the ECG. Only in this way is it possible to compare the
diagnostic performance of these methods. It is clear, of course, that if the source model is more accurate,
i.e., has more independent variables, the diagnostic performance is better, but when comparing ECG
and MCG, the comparison is relevant only if their complexity is similar [Malmivuo and Plonsey, 1995].

Detection of the Equivalent Magnetic Dipole of the Heart

The basic detection method of the equivalent magnetic dipole moment of a volume source is to measure
the magnetic field on each coordinate axis in the direction of that axis. To idealize the sensitivity
distribution throughout the volume source, the measurements must be made at a distance that is large
compared with the source dimensions. This, of course, decreases the signal amplitude. The quality of
the measurement may be increased considerably if bipolar measurements are used; i.e., measurements
are made on both sides of the source. Measurement of the magnetic field on each coordinate axis is,
however, difficult to perform in MCG due to the geometry of the human body. It would require either
six sequential measurements with one magnetometer (dewar) or six simultaneous measurements using
six dewars (Fig. 16.3).
It has been shown [Malmivuo, 1976] that all three components of the magnetic dipole also can be
measured from a single location. Applying this unipositional method symmetrically so that measurements
are made on both the anterior and posterior sides of the thorax at the same distance from the heart, only
two dewars are needed and a very high quality of lead fields is obtained. Figure 16.4 illustrates the
sensitivity distributions in nonsymmetrical and symmetrical measurements [Malmivuo and Plonsey,
1995].

FIGURE 16.3 Measurement of the three orthogonal components of the magnetic dipole moment of the heart (a) on
the coordinate axis (xyz lead system) and (b) at a single location over and under the chest (unipositional lead system).

© 2000 by CRC Press LLC

FIGURE 16.4 Sensitivity distributions in the measurement of the magnetic dipole moment of the heart.
(a) Nonsymmetrical and (b) symmetrical measurements of the x component. (c) Symmetrical measurement of the
y and z components.

Diagnostic Performance of ECG and MCG

The diagnostic performances of ECG and MCG were compared in an extensive study made at the Ragnar
Granit Institute [Oja, 1993]. The study was made using the asymmetrical unipositional lead system, i.e.,
making measurements only on the anterior side of the thorax. The patient material was selected, however,
so that myocardial changes were located dominantly on the anterior side.
This study consisted of 290 normal subjects and 259 patients with different myocardial disorders. It
was found that the diagnostic performance of ECG and MCG is about the same (83%). Diagnostic
parameters were then selected from both ECG and MCG. With this combined method, called electro-
magnetocardiogram (EMCG), a diagnostic performance of 90% was obtained. This improvement in
diagnostic performance was obtained without increasing the number of parameters used in the diagnostic
procedure. Moreover, this improvement is significant because it means that the number of incorrectly
diagnosed patients was reduced by approximately 50%.
This important result may be explained as follows: The lead system recording the electric dipole
moment of the volume source has three independent leads. (This is also the case in the 12-lead ECG

© 2000 by CRC Press LLC

 FIGURE 16.4 (continued)

system.) Similarly, the lead system detecting the magnetic dipole moment of the volume source has three
independent leads. Therefore, the diagnostic performances of these methods are about the same. However,
because the sensitivity distributions of electric and magnetic leads are different, the patient groups
diagnosed correctly with both methods are not identical.

© 2000 by CRC Press LLC

 FIGURE 16.4 (continued)

As stated before, the electric leads are independent of the magnetic leads. If the diagnostic procedure
simultaneously uses both the ECG and the MCG leads, we obtain 3 + 3 = 6 independent leads, and the
correctly diagnosed patient groups may be combined. Thus the diagnostic performance of the combined
method is better than that of either method alone. This is the first large-scale statistically relevant study
of the clinical diagnostic performance of biomagnetism.

© 2000 by CRC Press LLC

Technical Reasons to Use the MCG
The technical differences between ECG and MCG include the MCG’s far better ability to record static
sources, sources on the posterior side of the heart, monitor the fetal heart, and perform electrodeless
recording. As a technical drawback, it should be mentioned that the MCG instrument costs 2 to 3 times
more. An important feature of MCG is that, unlike the MEG instrument, it does not need a magnetically
shielded room. This is very important because the shielded room is not only very expensive but also
limits application of the technique to a certain laboratory space.

Theoretical Reasons to Use the MCG

It has been shown that MCG has clinical value and that it can be used either alone or in combination
with ECG as a new technique called the electromagnetocardiogram (EMCG). The diagnostic performance
of the combined method is better than that of either ECG or MCG alone. With the combined method,
the number of incorrectly diagnosed patients may be reduced by approximately 50%.

16.4 Magnetoencephalography (MEG)

Similarly as in the cardiac applications, in the magnetic measurement of the electric activity of the brain,
the benefits and drawbacks of the MEG can be divided into theoretical and technical ones. First, the
theoretical aspects are discussed.
The two main theoretical aspects in favor of MEG are that it is believed that because the skull is
transparent for magnetic fields, the MEG should be able to concentrate its measurement sensitivity in a
smaller region than the EEG, and that the sensitivity distribution of these methods are fundamentally
different. These questions are discussed in the following: The analysis is made using the classic spherical
head model introduced by Rush and Driscoll [Rush and Driscoll, 1969]. In this model, the head is
represented with three concentric spheres, where the outer radii of the scalp, skull, and brain are 92, 85,
and 80 mm, respectively. The resistivities of the scalp and the brain are 2.22 Ω·cm, and that of the skull
is 80 times higher, being 177 Ω·cm.
The two basic magnetometer constructions in use in MEG are axial and planar gradiometers. In the
former, both coils are coaxial, and in the latter, they are coplanar. The minimum distance of the coil
from the scalp in a superconducting magnetometer is about 20 mm. The coil radius is usually about 10
mm. It has been shown [Malmivuo and Plonsey, 1995] that with this measurement distance, decreasing
the coil radius does not change the distribution of the sensitivity in the brain region. In the following
the sensitivity distribution of these gradiometer constructions is discussed.
To indicate the magnetometer’s ability to concentrate its sensitivity to a small region, the concept of
half-sensitivity volume has been defined. This concept means the region in the source area (brain) where
the detector sensitivity is one-half or more from the maximum sensitivity. The smaller the half-sensitivity
volume, the better is the detector’s ability to focus its sensitivity to a small region.
In magnetocardiography, it is relevant to detect the magnetic dipole moment of the volume source of
the heart and to make the sensitivity distribution within the heart region as independent of the position
in the axial direction as possible. In magnetoencephalography, however, the primary purpose is to detect
the electric activity of the cortex and to localize the regions of certain activity.

Sensitivity Distribution of the Axial Magnetometer

In a cylindrically symmetrical volume conductor model, the lead field flow lines are concentric circles
and do not cut the discontinuity boundaries. Therefore, the sensitivity distribution in the brain area of
the spherical model equals that in an infinite, homogeneous volume conductor.
Figure 16.5 illustrates the sensitivity distribution of an axial magnetometer. The thin solid lines illustrates
the lead field flow lines. The dashed lines join the points where the sensitivity has the same value, being
thus so-called isosensitivity lines. The half-sensitivity volume is represented by the shaded region.

© 2000 by CRC Press LLC

 FIGURE 16.5 Sensitivity distribution of the axial magnetometer in measuring the MEG (spherical lead model).

Sensitivity Distribution of the Planar Gradiometer

Figure 16.6 illustrates the sensitivity distribution of a planar gradiometer. Again, the thin solid lines
illustrate the lead field flow lines, and the dashed lines represent the isosensitivity lines. The half-sensitivity
volume is represented by the shaded region. The sensitivity of the planar gradiometer is concentrated
under the center of the two coils and is mainly linearly oriented. Further, there exist two zero-sensitivity
lines.

Half-Sensitivity Volumes of Electro- and Magnetoencephalography Leads

The half-sensitivity volumes for different EEG and MEG leads as a function of electrode distance and
gradiometer baselines are shown in Fig. 16.7. The minimum half-sensitivity volume is, of course, achieved
with the shortest distance/baseline. For three- and two-electrode EEG leads, the half-sensitivity volumes
at 1 degree of electrode distance are 0.2 and 1.2 cm3, respectively. For 10-mm-radius planar and axial
gradiometer MEG leads, these volumes at 1 degree of coil separation (i.e., 1.6-mm baseline for axial
gradiometer) are 3.4 and 21.8 cm3, respectively.

© 2000 by CRC Press LLC

 FIGURE 16.6 Sensitivity distribution of the planar gradiometer (half-space model).

The 20-mm coil distance from scalp and 10-mm coil radii are realistic for the helmet-like whole-head
MEG detector. There exist, however, MEG devices for recording at a limited region where the coil distance
and the coil radii are on the order of 1 mm. Therefore, the half-sensitivity volumes for planar gradiometers
with 1-mm coil radius at 0- to 20-mm recording distances are also illustrated in Fig. 16.7. These curves
show that when the recording distance is about 12 mm and the distance/baseline is 1 mm, such a planar
gradiometer has about the same half-sensitivity volume as the two-electrode EEG.
Short separation will, of course, also decrease the signal amplitude. An optimal value is about
10 degrees of separation. Increasing the separation to 10 degrees increases the EEG and MEG signal
amplitudes to approximately 70 to 80% of their maximum value, but the half-sensitivity volumes do not
increase considerably from their values at 1 degree of separation.
Thus, contrary to general belief, the EEG has a better ability to focus its sensitivity to a small region
in the brain than the whole-head MEG. At about 20 to 30 degrees of separation, the two-electrode EEG
lead needs slightly smaller separation to achieve the same half-sensitivity volume as the planar gradiom-
eter. The sensitivity distributions of these leads are, however, very similar. Note that if the sensitivity
distributions of two different lead systems, whether they are electric or magnetic, are the same, they

© 2000 by CRC Press LLC

FIGURE 16.7 Half-sensitivity volumes of different EEG leads (dashed lines) and MEG leads (solid lines) as a function
of electrode distance and gradiometer baseline, respectively.

detect exactly the same source and produce exactly the same signal. Therefore, the planar gradiometer
and two-electrode EEG lead detect very similar source distributions.

Sensitivity of EEG and MEG to Radial and Tangential Sources

The three-electrode EEG has its maximum sensitivity under that electrode which forms the terminal
alone. This sensitivity is mainly directed radially to the spherical head model. With short electrode

© 2000 by CRC Press LLC

 distances, the sensitivity of the two-electrode EEG is directed mainly tangentially to the spherical head
model. Thus with the EEG it is possible to detect sources in all three orthogonal directions, i.e., in the
radial and in the two tangential directions, in relation to the spherical head model.
In the axial gradiometer MEG lead, the sensitivity is directed tangentially to the gradiometer symmetry
axis and thus also tangentially to the spherical head model. In the planar gradiometer, the sensitivity has
its maximum under the center of the coils and is directed mainly linearly and tangentially to the spherical
head model. The MEG lead fields are oriented tangentially to the spherical head model everywhere. This
may be easily understood by recognizing that the lead field current does not flow through the surface of
the head because no electrodes are used. Therefore, the MEG can only detect sources oriented in the two
tangential directions in relation to the spherical head model.

References
Baule GM, McFee R. 1963. Detection of the magnetic field of the heart. Am Heart J 55(7):95.
Cohen D. 1968. Magnetoencephalography: Evidence of magnetic fields produced by alpha-rhythm cur-
rents. Science 161:784.
Geselowitz DB. 1967. On bioelectric potentials in an inhomogeneous volume conductor. Biophys J 7(1):1.
Geselowitz DB. 1970. On the magnetic field generated outside an inhomogeneous volume conductor by
internal current sources. IEEE Trans Magn MAG-6(2):346.
Helmholtz HLF. 1853. Ueber einige Gesetze der Vertheilung elektrischer Ströme in körperlichen Leitern
mit Anwendung auf die thierisch-elektrischen Versuche. Ann Physik Chem 89:211.
Malmivuo J, Plonsey R. 1995. Bioelectromagnetism: Principles and Applications of Bioelectric and Bio-
magnetic Fields. New York, Oxford University Press.
Malmivuo JA. 1976. On the detection of the magnetic heart vector: An application of the reciprocity
theorem. Acta Polytechnol Scand 39:112.
Maxwell J. 1865. A dynamical theory of the electromagnetic field. Phil Trans R Soc (Lond) 155:459.
More PM, Feshbach H. 1953. Methods of Theoretical Physics, part I. New York, McGraw-Hill.
Oja OS. 1993. Vector Magnetocardiogram in Myocardial Disorders, MD thesis, University of Tampere,
Medical Faculty.
Örsted HC. 1820. Experimenta circa effectum conflictus electrici in acum magneticam. J F Chem Phys
29:275.
Plonsey R, Collin R. 1961. Principles and Applications of Electromagnetic Fields. New York, McGraw-Hill.
Rush S, Driscoll DA. 1969. EEG-electrode sensitivity: An application of reciprocity. IEEE Trans Biomed
Eng BME-16(1):15.
Zimmerman JE, Thiene P, Hardings J. 1970. Design and operation of stable rf biased superconducting
point-contact quantum devices. J Appl Phys 41:1572.

© 2000 by CRC Press LLC

View publication stats