Electroencephalography and clinical Neurophysiology, 86 (1993) 183-192 183

© 1993 Elsevier Scientific Publishers Ireland, Ltd. 0013-4649/93/$06.00

EEG 91139

Determination of current density distributions generated

by electrical stimulation of the human cerebral cortex
Surendar S. N a t h a n a , g Saurabh R. Sinha a, Barry G o r d o n b,d-g

Ronald P. Lesser b-d,f,g and Nitish V. Thakor a

Departments of a Biomedical Engineering, b Neurology, c Neurosurgery, d Cognitive Neurology, e Psychology, f Epilepsy Center,
and g Zanvyl Krieger Mind/Brain Institute, Johns Hopkins University, Baltimore, MD 21205 (USA)

( A c c e p t e d for publication: 24 August 1992)

Summary With the use of a 3-dimensional finite element model of the human brain based on structural data from MRI scans, we
simulated patterns of current flow in the cerebral hemisphere with different types of electrical stimulation. Five different tissue types were
incorporated into the model based on conductivities taken from the literature. The boundary value problem derived from Laplace's equation was
solved with a quasistatic approximation. Transcranial electrical stimulation with scalp electrodes was poorly focussed and required high levels of
current for stimulation of the cortex. Direct cortical stimulation with bipolar (adjacent) electrodes was found to be very effective in producing
localized current flows. Unipolar cortical stimulation (with a more distant reference electrode) produced higher current densities at the same
stimulating current as did bipolar stimulation, but stimulated a larger region of the cortex. With the simulated electrodes resting on the
pia-arachnoid, as usually occurs clinically, there was significant shunting of the current (7/8 of the total current) through the CSF. Possible
changes in electrodes and stimulation parameters that might improve stimulation procedures are considered.

Key words: Cerebral cortex; Current density distribution; Electrical stimulation; MRI scan; (Human)

Electrical stimulation of cerebral tissue, both direct
and transcranial, has been used for localization of
cortical functions and for medical diagnosis for many
years (Penfield and Jasper 1954; Merton and Morton
1980; Liiders et al. 1983; Lesser et al. 1984a,b,c; Roth-
well et al. 1987). Nevertheless, many questions remain
concerning electrode parameters, stimulating currents,
impedance of cerebral tissue, mode of stimulation
(transcranial, bipolar, or unipolar), and the regions
affected by stimulation. We attempted to answer some
of these questions by using a finite element simulation
of electrical stimulation of the human brain.
Our simulation differed in several major respects
from prior attempts. First, the model we used for the
brain was created from actual discretized MRI brain
images, rather than using idealized models such as
those assuming that the brain is a homogeneous vol-
ume conductor with a spherical geometry (Rush and
DriscoU 1968; Grandori and Rossini 1988; Roth et al.
1991). Our model incorporated the irregularity and
inhomogeneity of the brain and is therefore more
Correspondence to: Dr. Nitish V. Thakor, Dept. of Biomedical
Engineering, Johns Hopkins School of Medicine, 720 Rutland Ave.,
Baltimore, MD 21205 (USA).
accurate anatomically. Secondly, our simulation also
took into account the subdivisions of the cranial and
intracranial tissues (gray matter, white matter, and
CSF) and their differing resistivities. Thirdly, we used a
numerical method rather than an analytical method to
derive the results. The analytical methods that had
been used previously cannot yet be accurately applied
to the complex topographies that characterize the hu-
man brain. We therefore used a numerical technique,
the finite element method (FEM, Zienkiewicz 1977),
which has been shown to be robust and particularly
useful in situations involving curved boundaries and
complex boundary conditions (Walker et al. 1981;
Sepulveda et al. 1983; Kim et al. 1986; Fahy et al. 1988;
Kothiyal et al. 1988; Carter et al. 1989, 1990).
We compared the electrical field distributions in-
duced in the cerebral cortex by transcranial stimulation
with those produced by direct cortical stimulation. We
also compared the effects of adjacent cortical elec-
trodes ("bipolar") with those of more remote elec-
trodes ("unipolar"). In addition, we provide a detailed
analysis of the edge effects produced by the electrodes
typically used in subdural electrode grids for direct
cortical stimulation in humans (Lesser et al. 1984a,b,c;
Liiders et al. 1986; Gordon et al. 1990; etc.) and
suggest some ways these electrodes might be modified.
184 S.S. NATHAN ET AL.

Methods package uses the standard Gallerkin method to solve

the field problem.
Theory The ANSYS package can be used for our simulation
In finite element analysis, the domain under investi- if we treat temperature and heat flows as analogous to
gation (such as the brain) is considered to have a finite the voltages and currents of interest. The basic equa-
number of discontinuities in its properties and is di- tion for thermal energy balance is (ANSYS Theoretical
Manual 1989):
vided into a grid made of structures of mixed shape
called elements. The elements are chosen so as to fit
the overall shape of the object being modelled. The ~xx Kxx~-x + ~ y
0( 0(0W)
Kyy~-y +~zz Kzz~z +/~
elements are connected at nodes, which must be in the
corners of each element but may also be on the edges 0T
or in the center. Each of these elements is assigned a =pC-- (3)
at
value for relevant properties such as electrical conduc-
tivity. A list of nodes is assigned to the elements, and where K ~ , Kyy and Kzz are the thermal conductivities,
values for the unknown quantity in the differential /~ is the internal heat generation per unit volume, C is
equation are computed at these nodes. The values the specific heat, p is the density of the material, t is
between the nodes are interpolated. the time and T the temperature.
Our model characterizes volume conduction through Assuming quasistatic conditions (i.e., aT/at = 0), Eq.
(3) becomes
the brain. It is based on the assumption that the
medium consists only of conducting material. A quasi-
stationary condition is assumed for volume conduction, a [ aT
) { 0w)+0(0W)
a Kyy~y
~xx[K~'-0--xx + ~yy ~zz
Kzz-~-
z =0 (a)
that is, the time variations of the electric signal are not
considered. Heat conduction in one dimension is assumed to be
The electric field in a volume conductor is repre- governed by the Fourier law
sented as
aT
q = -k-- (5)
~Y. (~r IYV) = 0 (1) ax
O^ O^ 0^ where q is the heat flux, k is the conductivity, and x is
IF= - - i + '+--k the position. Analogously, the electric field problem is
Ox ~yyJ 0z
governed by Ohm's law
where V is the gradient vector, cr is the electrical
av
conductivity, V is the scalar electric potential and 1, ~,
J = - ~ 0---7 (6)
are unit vectors in the 3 planes.
Assuming uniform conductivity, from (1) we have where J is the current density, V is the scalar voltage
o2v O2V O2V 1 potential, or is the electrical conductivity, and x is the
~r. Ox----
T + -Oy
- 2 + 0---~-] = 0 position. Therefore, under quasistatic conditions, elec-
(21
tric field problems are solved in a manner similar to
which is the well known Laplace equation. The bound- heat transfer problems. The ANSYS software package
ary conditions are: (1) V = Vo (applied voltage at the solves heat transfer problems with general tempera-
point of contact of the electrode); (2) derivative of the ture-dependent conductivity; likewise, for electric field
scalar potential is zero at all other points. problems, solid body electric conduction with voltage-
The first condition is the Dirichlet boundary condi- dependent conductivity is modelled.
tion; the second is an example of the Neumann bound- For our simulation, the thermal conductivity was
ary condition. The Neumann boundary condition states replaced by the absolute magnitude of the complex
that no current flows in the direction normal to the electrical conductivity, the heat flux by the electrical
outer surface of the model, which is true since free current density, and the temperature by the electric
space is much more resistive to current flow than the potential. Thus, Eq. (4) can be rewritten as
tissue in the brain. Eq. (2) is numerically solvable for
regions with regular geometries. For regions with com-
0t0v I
~xx~ ~-x]+~yl%Y-~--y
0v) 0{ 0v)
+~zz Crzz~z = 0 (7)
plex geometries like the brain, analytical solution is
impossible, and the finite element method can be used. If the medium is isotropic (i.e., ~rx~= crrj = O'zz= cr),
We used ANSYS (Swanson Analysis Inc., Houston, the above equation reduces to Laplace's equation. AN-
PA), a commercial FEM package, for this purpose. SYS uses a 2 x 2 × 2 lattice of integration points for
This package is set to generate temperatures and heat solving Laplace's equation by the gaussian integration
flows at the nodes (ANSYS Users Manual 1989). The procedure (ANSYS Theoretical Manual 1989).
C U R R E N T DENSITY DISTRIBUTIONS IN ELECTRICAL STIMULATION
TABLE I
The resistivities used in the model.
Tissue Electrical resistivity
(O.cm)
Scalp 1000
Skull 8 000
Gray matter 250
White matter 500
CSF 65
The electric field can be derived from the scalar
potential as
E = - ffV (8)
The current density can be calculated from the
electric field as
J = or. E (9)
where J is the current density, tr is the conductivity,
and E is the electric field.
Finite element model
M R I scans of the horizontal section through the
long axis of a normal brain were used to build the
model. The original M R I image was represented by
256 x 256 pixels. A 4-neighborhood average was done
on the image to reduce the resolution to a size that
could be accommodated by the F E M solver. Based on
the histogram of the averaged image, boundaries were
set and the pixels between boundaries were assigned as
5 different materials: scalp, skull, gray matter, white
matter, and CSF. Cubic elements were built from these
pixels and the resistivities (Table I) were taken from
standard sources (Geddes and Baker 1967; Geddes
1987). The same process was repeated for 9 slices in
order to create a 3-dimensional (3-D) model.
Brain tissues are known to be anisotropic (Nichol-
son 1965; Foster et al. 1979). However, the level of
resolution and anatomical detail from the M R I scans
did not permit us to investigate anisotropy. Therefore,
the head as a whole was treated as an inhomogeneous
medium containing alternating layers of isotropic re-
gions. Transcranial stimulation was simulated by using
the 3-D model of the brain. The elements in the model
were cubical (0.6 x 0.6 x 0.6 cm). The scalp electrodes
used in the model were rectangular (1.8 x 0.6 cm), as
have been used clinically (Levy et al. 1984). Stimulation
was carried out via two electrodes (driven and ground).
The ground electrode was placed in the same position
for all the simulations while the driven electrode was
moved to obtain results with various electrode separa-
tions. In all the simulations, a stimulating current of
300 m A (to reproduce real situations) was used. The
185
electrodes were placed in the middle of the stack of
the 3-D model (the fifth slice).
To simulate stimulation with the subdural electrode
grid, we isolated the elements representing the skull
and the scalp from the 3-D brain model. A 6 x 6
electrode grid similar to the 8 x 8 grid normally used
(Lesser et al. 1987) was placed on the cortex in the
model. The electrode center-to-center separation was
1.0 cm, and the diameter of each electrode was 0.3 cm.
These values are identical to those of the arrays cur=
rently used at this institution. We performed two kinds
of simulations with this electrode grid: for brevity, we
used the term "unipolar" for stimulation between a
subdural electrode within the region of interest and a
remote electrode and "bipolar," for stimulation be-
tween adjacent subdural electrodes.
The direct cortical simulations used a detailed 2-D
model of the brain for improved anatomical accuracy.
This was built from a single M R I scan (Fig. la), with
different layers in the model representing the CSF,
gray matter, and white matters. Fig. lb shows a 3 cm
region of the model enlarged to give a better view of
the region surrounding the electrodes.
2-D models depicting the region surrounding the
electrodes were built to investigate the current density
profiles under the electrode surface. Disk and hemi-
spherical types of electrodes were modelled and their
dimensions are shown in Fig. 2a and b.
3-D disk electrode models were also built, to study
the effect of recession of the surface of the electrodes
on the underlying patterns of current density. The
models contained all the 5 tissue types (Fig. 2c) and the
parameters used are listed in Table III.
Results of cortical stimulation
Transcranial stimulation
Transcranial stimulation with the use of scalp elec-
trodes is intrinsically appealing because it is non-inva-
sive. It has been used for mapping the motor system
(Cohen et al. 1987). However, transcranial stimulation
by scalp electrodes does have major limitations, which
were confirmed by our simulation. Since resistivity of
the skull is high relative to that of the scalp, most of
the current gets shunted through the scalp (Fig. 3).
This limits the amount of current that is conveyed to
the cerebral tissue. Since the quasistatic field approxi-
mation implies conservation and linearity of the elec-
tric field solution, different electric field and current
density values can be linearly extrapolated from the
above result obtained with a 300 m A source. The
maximum current density in the scalp occurs at an
electrode separation of 1.8 cm. More insight into mod-
els of transcranial stimlation can be found in the works
of Rush and Driscoll (1968) and Grandori and Rossini
186 S.S. N A T H A N E T AL.

F
0.3 cn] ~ ~ 0.3 cnl -~

matter (a) (b)

ite matter

Stimulat
electrodt

(a) Non-corn
layer
aaller

CSF
Fig. 2. Electrode dimensions: (a) the hemispherical electrodes; (b)
the disk electrodes; and (c) the 3-D model of brain tissue.
m
White Matter

studied by Ojemann and his co-workers (Whitaker and

Ojemann 1977) who used bipolar silver ball electrodes
G r a y Matter for stimulation.
More recently, indwelling subdural electrode arrays
and strips have been used to apply electrical stimula-
CSF !
Electrodes
7"

(b) ~ Scalp
Fig. 1.2-D brain model, a: 2-dimensional model of the brain created
~ 6-
Gray Matter
with a mesh of 4024 elements. Taking advantage of symmetry, only
half the brain is modelled. CSF, gray matter and white matter are
shown, b: 3 cm of the 2-dimensional model of the brain enlarged for
~ 5-
better view around the stimulating electrodes and to show the shape
of the sulci and the gyri. T h e n u m b e r of elements is 2563.
~ 4-

~ 3-
(1988). As the simulation helps confirm, focussing of
stimulation by scalp electrodes is very difficult. There-
fore, our work focusses primarily on direct cortical 2-
stimulation.

Stimulation with cortical electrodes 0 4 8 12 16

Direct cortical electrical stimulation to localize lan-
guage areas was established by Penfield and his collab-
orators (Penfield and Jasper 1954; Penfield and Roberts
1959; Penfield and Perot 1963). They used ball elec-
trodes placed on the cortical surface. Functional local-
ization of the cortex by electrical stimulation was also
Electrode separation (cm)
Fig. 3. Transcranial stimulation: the peak current densities in the
various tissues of the brain induced by transcranial stimulation by a
300 m A source with scalp electrodes. T h e plot shows the current
densities in the slice containing the electrodes for different electrode
separations.
CURRENT DENSITY DISTRIBUTIONS IN ELECTRICAL STIMULATION 187

tion directly to the cortex to localize motor, sensory,
speech and other functional cortical regions (Lueders
et al. 1983; Wyler et al. 1984; Lesser et al. 1987; Liiders
et al. 1989). Typically, electrode grids contain 4 x 4 or
8 x 8 electrodes (Li~ders et al. 1989), although other
configurations have also been used, and strips are
typically 1 x 4 to 1 x 8. The spacing between the elec-
trodes is usually 1 cm. Individual electrodes usually
have had 2-3 mm of exposed surface.
We simulated the two basic forms of cortical electri-
cal stimulation, discussed earlier (bipolar and unipolar)
and studied the spread of current density laterally as

0.0500
0.0450
0.04OO
Gray Matter 0.0350
0.0300
0.0250
0.0200
0.0150
White Matter 0.0100
0.0050
0.0000

5 mm

0.014

0.012

0.010

0.008

0.006

0.004

0.002

0.000
(a) (b)
Fig. 4 (Top). Bipolar stimulation: spread of current density laterally as well as with depth. The current densities ( A / c m 2) generated in the cortex
by bipolar stimulation. The two tissue types shown are the gray and white matters. The position of the electrodes are shown in the figure.

Fig. 5 (Bottom). Unipolar stimulation: the current density distribution immediately under the subdural electrode grid. Panels a and b represent
the current densities at depths of 2 and 10 mm under the electrode grid. The reference electrode is in the corner of the grid. The distance
between the electrodes is 8 cm.
188 S.S. N A T H A N E T AL.

well as at various depths in the cortex. The current T A B L E II

densities immediately under the electrode grid were M a x i m u m and m i n i m u m current densities in the region close to the
represented as a 3-D surface plot. electrodes *.
Bipolar stimulation. Fig. 4 shows the current densi- Tissue Max current density Min current density
ties in the cortex for bipolar stimulation with 10 mA ( m A / c m 2) ( m A / c m 2)
stimulating current. The peak current density occurs in CSF 49.09 5.49
the region immediately beneath the bipolar electrodes Gray matter 5.13 0.73
(0.05 A / c m 2) and it declines rapidly to 0.02 A / c m 2, White matter 1.52 0.18
0.5 cm away. The current density decreases in relation * The stimulating current was 2 mA.
to the square of the distance into the cortex. The
spread of the current density laterally as well as with
T A B L E III
depth is clearly brought out in Fig. 4.
3-D electrode model parameters.
Unipolar vs. bipolar stimulation. Fig. 5 shows the
current density in the region between the stimulating Model parameters Values (cm)
unipolar electrodes for depths of 0.2 and 1.0 cm re- Thickness of white matter 1
spectively. With unipolar stimulation, the current den- Thickness of gray matter 0.5
sity decreases much less rapidly with depth (Fig. 5a and Thickness of CSF 0.1
Recession 0.04
b). Unipolar stimulation also tends to stimulate a wider
Electrode separation 1, 2
region than does bipolar stimulation. Exposed electrode radius 0.158
Increasing the stimulating currents results in in- Total electrode radius 0.198
creased current densities; this is represented in Fig. 6. Electrode thickness 0.0025
We can see that as the stimulating current decreases,
the peak value of the current density also decreases.
Shunting. Gordon et al. (1990) suggested that, ow- does provide an effective shunting path for the cur-
ing to the thicker pia arachnoid, a larger volume of rents.
CSF, and thicker superficial layers in the human cor-
tex, there is a greater distance between the electrode
and cortical tissue, and hence more opportunity for Electrode analysis
current shunting through the CSF in humans than
there is in the CSF of primates. We tested this hypoth- Consequences of the geometrical design of the elec-
esis in our model (Table II). As predicted, the CSF trodes can be easily determined with our simulation.
Low electrode-body interface impedance causes "edge
effect," that is, an increase in the current density
towards the perimeter of the electrode (Lindemanns et
al. 1975; Kim et al. 1984). Such non-uniformity can
1.50 potentially cause problems such as focal tissue damage
o4
or pain. Different electrode designs have therefore
Sthnulathag Current been studied in attempts to create a more uniform
1.20
-.....o...- 5mA distribution of current density under the electrode
--<>--- 3rnA surface (Wiley and Webster 1982; Kim et al. 1984,
0.90 \ ........~........ lmA 1986).
o
The simulated current density profile under the
surface of the disk electrodes is shown in Fig. 7a, which
0.60 includes a schematic of the model itself in the inset.
The electrode separation is 1.0 cm. The edge effect
phenomenon can be clearly seen (see Fig. 7b, which
0.30 shows the current density on the surface of the disk
electrodes for electrode separations of 0.5 and 1.0 cm).
As suggested by Wiley and Webster (1982), we as-
0.00 . . .
sessed an electrode that generates a non-uniform po-
0.00 1.00 2.00 3.00
tential distribution to obtain uniform current density
Distance from surface o f Cortex (cm) distributions across the electrode surface. The same
Fig. 6. Line graph of the average current densities over a 2 cm region simulations were therefore carried out with hemispher-
surrounding the stimulating electrodes. Bipolar stimulation: the peak
current density in a spherical region of radius 2 cm around the
ical electrodes with a diameter of 0.3 cm. Fig. 8a shows
bipolar electrodes in the subdurai electrode grid at various depths the resulting current densities under the electrode sur-
from the surface of the cortex and at 3 different stimulating currents. face along with the brief schematic of the model. The
C U R R E N T D E N S I T Y D I S T R I B U T I O N S IN E L E C T R I C A L S T I M U L A T I O N 189

edge effect is considerably reduced. The current densi- 0.08

ties under the different types of electrodes (disk and
0.07 ~'///~ White matter
hemispherical) are compared in Fig. 9, with the region
near the driven electrode enlarged to show the profile Gray matter
eq
0.06
I ] CSF
of a single electrode. The current density is more E
0.05. Electrodes
uniform across the surface of the hemispherical elec-
trode than it is for the disk electrode.
0.04.
Fig. 8b shows the current densities on the surface of
the brain for hemispherical electrodes at separations of 0.03.
0.5 and 1 cm. The current density values are lower than
those for disk electrodes (Fig. 7b) with the same stimu- 0.02.

lation parameters. The current densities at electrode 0.01

separations of greater than or equal to 1 cm are also
relatively more uniform with the hemispherical elec- 0,(30 i i | i i i i

trodes than with the disk electrodes. The values of 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
current densities at electrode separations of greater Electrode axis (cm)

(a)
0.08
0.1(X) _

} / / / / A White Matter Electrode

Gray Matter A separation
(era)
I I CSF
0.075 0.06 0.5
Electrodes
.<

,~ 0.050 "~ 0.O4.

~ 0.025 I.
= 0.02.

0.000 | i
I_ _1 i i
i i i i i J O.(X) i i i w i i i

0.00 0.40 0.80 1,20 1,60 2,00 2,40 0.0 (1.4 0.8 1.2 1.6 2.0 2.4 2.8
Electrode axis (cm) Electrode axis (cm)
(a)
(b)
0.10 Fig. 8. Hemispherical electrodes, a: The current densities under the
surface of the hemispherical electrodes. The electrode model is
Electrode shown in the inset (b). The current densities under the surface of the
0.08 separation (cm) electrodes for electrode separations of 0.5 and 1 cm.
0.5
1.0
~,~ 0.06 than or equal to 1 cm are almost the same for hemi-
.,~ spherical and disk electrodes. The important differ-
ences between these two electrode types occur at stim-
'~ 0.04 ulation depths of less than 3 cm; at greater depths, the
current densities are virtually identical for the two
I,N
types of electrode.
(,.) 0.02,
In order to replicate the exact parameters of elec-
trodes used in electrical stimulation of the cerebral
0.00 i i i i
cortex at our institution (Gordon et al. 1990), we built
0.00 0.60 1.20 1.80 2.40 a 3-D model of a disk electrode, recessed 0.04 cm
Electrode Axis (era) below the silastic surface of the array, and a multilayer
brain (Fig. 2c) and studied the effect of recessing the
(b)
electrodes with a resistive material. The edge effect is
Fig. 7. Disk electrodes, a: the current densities under the surface of
the disk electrodes. The electrode model is shown in the inset (b). virtually non-existent in the case of recessed electrodes
The current densities on the surface of the same disk electrodes for (Fig. 10). We also studied the effect of different elec-
electrode separations of 0.5 and 1 cm. trode separations using this model. When the electrode
190 S.S. NATHAN ET AL.

0.05'

Disc
Hemispherical
~" 0.04' 80
Current 60
density 2 40
.<
(mA/cm) 2c
(
"~ 0.03"

v
30
0.02

(a)
0.01 I , i ,
0.25 0.45 0.65 0.85
E l e c t r o d e Axis (em)
Fig. 9. Edge effect: the current densities under the surface of the
driven electrodes (disk and hemispherical). The current density is
almost uniform along the surface of the hemispherical electrodes as 1
compared to that of disk electrodes. Current 1
densiw 2
(mA/cm)
separation was doubled, the maximum value of the
current density dropped (Fig. 11a), whereas when the
electrode separation was halved, the current density
maxima increased (Fig. llb).
30

Discussion (b)
Fig. 11. Recessed disk electrodes with varied electrode separation.
Our simulation has addressed several issues of theo- The current density distribution under the surface of the recessed
retical and practical importance in performing and electrodes in the 3-D electrode model for different electrode separa-
tions (a) 2 cm and (b) 0.5 cm. The current density is higher with a
separation of 0.5 cm as compared to that of a 2 cm separation.
1.0

interpreting cerebral electrical stimulation studies. The

~
¢q
. 0.8 main objective of these simulations was to probe into
E the field distributions generated in the cerebral cortex
.< by electrical cortical stimulation. Transcranial electri-
"-" 0.6 cal stimulation is a relatively poor, unfocussed way of
delivering current to the cerebral tissue, compared to
direct cortical stimulation. Direct cortical electrical
0.4
stimulation is much more efficient and focussed. Our
t_
t_ s simulations show that bipolar simulations produce a
higher current density than unipolar stimulation in a
0.2 / ~
non recessed
small region surrounding the electrodes. We have also
confirmed that outside this region, unipolar stimulation
recessed
0.0 ....
tends to produce a higher current density than bipolar
0.0 0.4 0.8 1.2 1.6 2.0 stimulation (because unipolar stimulation is actually
bipolar, but to a more distant set of electrodes) (see
Electrode axis (cm)
Figs. 9-11). Therefore, bipolar stimulation tends to
Fig. 10. Non-recessed vs. recessed disk electrodes. The current
produce a more focussed region of electrical stimula-
density distribution under the surface of the non-recessed and re-
cessed electrodes in the 3-D electrode model with a stimulating tion. These inferences were also made by Stark et al.
current of 10 mA. The electrode separation is 1 cm. (1962) and Phillips and Porter (1962).
CURRENT DENSITY DISTRIBUTIONS IN ELECTRICAL STIMULATION
We explored several possible configurations of bipo-
lar electrodes. Maximal focus of current density can
probably be better achieved with a 0.5 cm inter-elec-
trode separation than with the 1 cm inter-electrode
separation that is currently used in clinical practice.
Edge effects are a potential problem of surface grid
electrodes. These edge effects, however, seem to be
considerably reduced by the usual recessing of the
surface of these electrodes. They can be further re-
duced by decreasing the inter-electrode separation,
and by changing the shape of the electrode to hemi-
spherical or spherical rather than flat disk electrodes.
These variations might be considered in the future
design of cortical stimulation electrodes.
These simulations have also provided us with the
necessary foundation for comparing the field distribu-
tions produced by electrical stimulation with those
produced by the increasingly popular transcranial mag-
netic stimulation.
This work was supported in part by NIH (NINDS) 1-RO1-
NS26553, NIH (NINDS) I-RO1-NS24282, by grants from the Seaver
Institute, L.K. Whittier Foundation, Pittsburgh Supercomputing
Center through the NIH Division of Research Resources Coopera-
tive Agreement 1 P41 RR06009-01, and the McDonnell Pew Pro-
gram in Cognitive NeuroScience.
We thank Swanson Analysis Systems, Inc., for the use of ANSYS
software under an academic license.
We also wish to thank Dr. Pamela Talalay for editorial support.
References
ANSYS Engineering Analysis System Theoretical Manual. Swanson
Analysis Systems, Houston, TX, 1989.
ANSYS Engineering Analysis System User's Manual. Swanson Anal-
ysis Systems, Houston, TX, 1989.
Carter, E.L., Vresilovic, E.J., Pollack, S.R. and Brighton, C.T. Field
distributions in vertebral bodies of the rat during electrical stimu-
lation: a parametric study. IEEE Trans. Biomed. Eng., 1989, 36:
333-345.
Carter, E.R., Pollack, S.R. and Brighton, C.T. Theoretical determi-
nation of the current density distributions in human vertebral
bodies during electrical stimulation. 1EEE. Trans. Biomed. Eng.,
1990, 37: 606-614.
Cohen, L.G., Hallett, M. and Johnson, L. Non-invasive mapping of
the human motor cortex. Neurology, 1987, 37: 219-221.
Fahy, J.B., Kim, Y. and Ananthaswamy, A. Optimal electrode config-
urations for external cardiac pacing and defibrillation: an inho-
mogeneous study. IEEE Trans. Biomed. Eng., 1987, 34: 743-748.
Foster, K.R., Schepps, J.L., Stoy, R.D. and Schwan, H.P. Dielectric
properties of brain tissue between 0.01 and 10 GHz. Phys. Med.
Biol., 1979, 24: 1177-1187.
Geddes, L.A. Optimal stimulus duration for extracranial cortical
stimulation. Neurosurgery, 1987, 20: 94-99.
Geddes, L.A. and Baker, L.E. The specific resistance of biological
material - a compendium of data for the biomedical engineer
and physiologist. Med. Biol. Eng., 1967, 5: 271-293.
Gordon, B., Lesser, R.P., Rance, N.E., Hart, J., Webber, R., Ue-
matsu, S. and Fisher, R.S. Parameters for direct cortical electrical
stimulation in the human: histopathologic confirmation. Elec-
troenceph, clin. Neurophysiol., 1990, 75: 371-377.
191
Grandori, F. and Rossini, P. Electrical stimulation of the motor
cortex: theoretical consideration. Ann. Biomed. Eng., 1988, 16:
639-652.
Kim, Y., Webster, J.G. and Tomkins, W.J. Simulated and experimen-
tal studies of temperature elevation around electrosurgical dis-
persive electrodes. IEEE Trans. Biomed. Eng., 1984, 31: 681-692.
Kim, Y., Fahy, J.B. and Tupper, B.J. Optimal electrode designs for
electrosurgery, defibrillation and external cardiac pacing. IEEE
Trans. Biomed. Eng., 1986, 33: 845-853.
Kothiyal, K.P., Shankar, B., Fogelson, L.J. and Thakor, N.V. Three
dimensional computer model of electric fields in internal defibril-
lation. Proc. IEEE, 1988, 76: 720-730.
Lesser, R.P., Lueders, H., Dinner, D.S. et al. The location of speech
and writing functions in the frontal language area; results of
extraoperative cortical stimulation. Brain, 1984a, 107: 275-291.
Lesser, R.P., Dinner, D.S., Lueders, H. et al. Differential diagnosis
and treatment of intractable seizures. Cleve. Clin. Quart., 1984b,
51: 227-240.
Lesser, R.P., Lueders, H., Klem, G. et al. Cortical afterdischarge and
functional response thresholds: results of extraoperative testing.
Epilepsia, 1984c, 25: 615-621.
Lesser, R.P., Liiders, H., Klem, G., Dinner, D.S., Morris, H.H.,
Hahn, J.F. and Wyllie, E. Extraoperative cortical functional local-
ization in patients with epilepsy. J. Clin. Neurophysiol., 1987, 4:
27-53.
Levy, W.J., York, D.H., McCaffrey, M. and Tanzer, F. Motor evoked
potentials from transcranial stimulation of the motor cortex in
humans. Neurosurgery, 1984, 15:287-302
Lindemanns, F.W., Heethaar, R.M., Denier van der Gon, J.J. and
Zimmerman, A.N.E. Site of initial excitation and current thresh-
old as a function of electrode radius in heart muscle. Cardiac
Res., 1975, 9: 95-104.
Liiders, H., Lesser, R.P., Dinner, D.S. et al. The second sensory area
in man: evoked potential and electrical stimulation studies. Neu-
rology, 1983, 33 (Suppl. 2): 185.
Liiders, H., Lesser, R.P., Hahn, J., Dinner, D.S., Morris, H., Sesor,
S. and Harrison, M. Basal temporal language area demonstrated
by electrical stimulation. Neurology, 1986, 36: 505-510.
LiJders, H., Hahn, J., Lesser, R.P., Dinner, D.S., Morris, H.M.,
Wyllie, E., Friedman, L. and Skipper, G. Basal temporal subdural
electrodes in the evaluation of patients with intractable epilepsy.
Epilepsia, 1989, 30:131 142.
Merton, P.A. and Morton, H.B. Stimulation of the cerebral cortex in
the intact human subject. Nature, 1980, 285: 227.
Nicholson, P.W. Specific impedance of cerebral white matter. Exp.
Neurol., 1965, 13: 386-401.
Ojemann, G.A. Brain organization for language from the perspective
of electrical stimulation mapping. Behav. Brain Sci., 1983a, 6:
189-230.
Ojemann, G.A. Electrical stimulation and the neurobiology of lan-
guage. Behav. Brain Sci., 1983b, 6: 221-226.
Penfield, W. and Jasper, H. Epilepsy and the Functional Anatomy of
the Human Brain. Little Brown, Boston, MA, 1954.
Penfield, W. and Perot, P. The brain's record of auditory and visual
experience - a final summary and discussion. Brain, 1963, 86:
595 -696.
Penfield, W. and Roberts, L. Speech and Brain Mechanisms. Prince-
ton University Press, Princeton, NJ, 1959.
Phillips, C.G. and Porter, R. Unifocal and bifocal stimulation of the
motor cortex. J. Physiol. (Lond.), 1962, 162: 532-538.
Roth, B.J., Saypol, J.M., Hallett, M. and Cohen, L.G. A theoretical
calculation of the electric field induced in the cortex during
magnetic stimulation. Electroenceph. clin. Neurophysiol., 1991,
81:47-56
Rothwell, J.C., Thompson, P.D., Day, B.L., Dick, J.P.R., Kachi, T.,
Cowan, M.A. and Marsden, C.D. Motor Cortex stimulation in
192
intact man. 1. General characteristics of EMG responses in
different muscles. Brain, 1987, 110: 1173-1190.
Rush, S. and Driscoll, D.A. Current distribution in the brain from
surface electrodes. Anesth. Analg., 1968, 47: 717-723.
Sepulveda, N.G., Walker, C.F. and Heath, R.G. Finite element
analysis of current pathways with implanted electrodes. J. Biomed.
Eng., 1983, 5: 41-48.
Stark, P., Fazio, G. and Boyd, E.S. Monopolar and bipolar stimula-
tion of the brain. Am. J. Physiol., 1962, 203: 371-373.
Walker, C.F., Sepulveda, N.G. and Rusinko, J.B. A new technique
for stimulation electrode design. IRCS Med. Sci., 1981, 9: 681-
685.
S.S. NATHAN ET AL.
Whitaker, H.A. and Ojemann, G.A. Graded localization of naming
from electrical stimulation mapping of left cerebral cortex. Na-
ture, 1977, 270: 50-51.
Wiley, J.D. and Webster, J.G. Analysis and control of the current
distribution under circular dispersive electrodes. IEEE. Trans.
Biomed. Eng., 1982, 29: 381-385.
Wyler, A.R., Ojemann, G.A., Lettich, L. and Ward, A.A. Subdural
strip electrodes for localizing epileptogenic foci. J. Neurosurg.,
1984, 60: 1195-1200.
Zienkiewicz, O.C. The Finite Element Method. McGraw Hill, New
York, 1977.