32nd Annual International Conference of the IEEE EMBS
Buenos Aires, Argentina, August 31 - September 4, 2010
Abstract—The goal of this study is to investigate the regional
distribution of the electric field (E-field) strength induced by
electroconvulsive therapy (ECT), and to contrast clinically
relevant electrode configurations through finite element (FE)
analysis. An FE human head model incorporating tissue hete-
rogeneity and white matter anisotropy was generated based on
structural magnetic resonance imaging (MRI) and diffusion
tensor MRI (DT-MRI) data. We simulated the E-field spatial
distributions of three standard ECT electrode placements [bi-
lateral (BL), bifrontal (BF), and right unilateral (RUL)] and an
investigational electrode configuration [focal electrically admi-
nistered seizure therapy (FEAST)]. A quantitative comparison
of the E-field strength was subsequently carried out in various
brain regions of interests (ROIs) that have putative role in the
therapeutic action and/or adverse side effects of ECT. This
study illustrates how the realistic FE head model provides
quantitative insight in the biophysics of ECT, which may shed
light on the differential clinical outcomes seen with various
forms of ECT, and may guide the development of novel stimu-
lation paradigms with improved risk/benefit ratio.
LECTROCONVULSIVE therapy (ECT) is a highly
effective medical treatment in which electric currents
applied through electrodes on the scalp of anesthetized pa-
tients induce a generalized seizure. Although ECT plays a
vital role in the treatment of medication-resistant psychiatric
disorders, such as major depression, there is still limited
knowledge on how to optimally select electrode placement or
stimulus current parameters. For example, the efficacy and
Manuscript received April 23, 2010; revised June 25, 2010. This work
was supported in part by the National Science Foundation through TeraGrid
resources provided by NCSA under grant number TG-MCB100050 and by
NIH NCRR (5TL1RR024158-03). *These authors contributed equally to this
W. H. Lee is with the Department of Biomedical Engineering and with the
Division of Brain Stimulation and Therapeutic Modulation, Department of
Psychiatry, Columbia University, New York, NY 10032, USA (e-mail:
Z.-D. Deng and A. V. Peterchev are with the Department of Electrical
Engineering and with the Division of Brain Stimulation and Therapeutic
Modulation, Department of Psychiatry, Columbia University, New York, NY
10032, USA (phone: 212-543-5460; fax: 212-543-4284; e-mail:
T.-S. Kim is with the Department of Bioemdical Engineering, Kyung Hee
University, Republic of Korea (e-mail: email@example.com).
A. F. Laine is with the Department of Biomedical Engineering, Columbia
University, New York, NY 10027, USA (e-mail: firstname.lastname@example.org).
S. H. Lisanby is with the Division of Brain Stimulation and Therapeutic
Modulation, Department of Psychiatry, Columbia University / New York
State Psychiatric Institute, New York, NY 10032, USA (e-mail:
side effects of ECT are highly dependent upon electrode
placement, but a complete biophysical explanation for these
relationships is lacking. For instance, right unilateral (RUL)
has fewer cognitive side effects than bilateral (BL) ECT, but
it is not known whether this is by virtue of lower electric field
(E-field) strength in hippocampus and in other regions im-
portant for memory. Furthermore, certain electrode configu-
rations such as bifrontal (BF) and focal electrically adminis-
tered seizure therapy (FEAST) have been proposed with the
goal of preferentially targeting prefrontal cortex , but their
relative E-field strengths in frontal regions have not been
quantified. Previously, using a spherical head model, we
found that BL induces stronger E-field and stimulates a
higher percentage of total brain volume compared to RUL
and FEAST , . However, the spherical model has limi-
tations as it cannot account for the complex tissue geometries
of the head, anisotropic tissue properties, and orifices in the
A number of studies have used computational head models
with realistic shape to examine the effects of ECT –.
However, there have been no modeling studies that compare
various forms of ECT by quantifying the E-field in specific
brain regions of interest (ROIs).
In this study we use an anatomically-accurate finite ele-
ment (FE) human head model incorporating white matter
(WM) anisotropy to quantify regional differences in E-field
strength induced by various ECT electrode configurations.
The FE model is derived from individual structural magnetic
resonance imaging (MRI) and diffusion tensor MRI
(DT-MRI) scans of a human head. The WM anisotropic
conductivity is estimated from the DT-MRI. This model
allows us to investigate whether forms of ECT associated
with fewer cognitive side effects induce lower E-field
strengths in hippocampus, and to evaluate the degree to
which frontal electrode configurations achieve focal frontal
stimulation. This study demonstrates the utility of anatomi-
cally-realistic computational models to provide clinically
salient analysis and recommendations for the optimization of
The steps of the E-field modeling and ROI analysis are
diagrammed in Fig. 1 and described below.
A. FE Mesh Generation
To build the realistic head model, we used an adaptive
Won Hee Lee*, Student Member, IEEE, Zhi-De Deng*, Student Member, IEEE, Tae-Seong Kim, Member, IEEE,
Andrew F. Laine, Fellow, IEEE, Sarah H. Lisanby, and Angel V. Peterchev, Member, IEEE
Regional Electric Field Induced by Electroconvulsive Therapy:
A Finite Element Simulation Study
978-1-4244-4124-2/10/$25.00 ©2010 IEEE2045
meshing technique that generates DT-MRI anisotro-
py-adaptive FE meshes . This mesh generation technique
produces FE meshes adapted to the anatomical details of the
head, including tissue boundaries derived from structural
MRIs and WM anisotropy from measured diffusion tensors.
The steps of the adaptive meshing scheme are summarized as
follows: (i) Structural feature maps are derived from the
anatomical MRIs using the structure tensor . (ii) Initial
MRI content-adaptive FE nodes are sampled based on the
spatial density of the feature maps. (iii) A map of fractional
anisotropy as an anisotropy feature map is calculated using
the eigenvalues of the diffusion tensor matrix. (iv) WM ani-
sotropy-adaptive FE nodes in the WM tissues are extracted
based on the spatial anisotropic density of the WM fractional
anisotropy maps. (v) Final FE meshes are created using the
Delaunay tessellation algorithm. The detailed methodology
of generating WM anisotropy-adaptive FE meshes is de-
scribed in . The resulting FE model of the human head
consists of five sub-regions including WM, gray matter,
cerebrospinal fluid, skull, and scalp, incorporating a total of
160,230 nodes and 1,009,440 tetrahedral elements.
B. Tissue Properties
All tissues except WM were assumed to be isotropic .
The isotropic conductivities were set to 0.33 S/m for gray
matter, 1.79 S/m for cerebrospinal fluid, 0.0132 S/m for
skull, and 0.35 S/m for scalp . To estimate the WM
conductivity tensors, we used the assumption that the con-
ductivity tensors share eigenvectors with the measured dif-
fusion tensors . Then we deployed the volume constraint
algorithm with a fixed anisotropic ratio of 10:1 in each WM
voxel, yielding electrical conductivity estimates of 0.65 S/m
and 0.065 S/m in the longitudinal (parallel) and transverse
(perpendicular) WM fiber orientation, respectively .
C. ECT Electrode Configurations
Three conventional ECT electrode placements (BL, BF,
and RUL) and an investigational configuration (FEAST )
were modeled (see Fig. 2). For BL ECT, two electrodes were
placed at the frontotemporal positions. For BF ECT, elec-
trodes were placed 5 cm above the outer angle of the orbit on
a line parallel to the sagittal plane. For RUL ECT, one elec-
trode was placed to the right of vertex and the second elec-
trode was in the right frontotemporal position. For FEAST,
one electrode was placed over the right motor strip and the
second electrode was placed over the right eyebrow . The
ECT electrodes were approximated as point sources placed at
the centers of the electrodes shown in Fig. 2.
D. ECT Simulation
The E-field distribution for all ECT electrode configura-
tions was simulated at a current of 800 mA using the com-
mercial software ANSYS (ANSYS Inc., Canonsburg, PA).
Since the frequencies used in ECT are relatively low, the
E-field solution was obtained by solving the quasi-static
Laplace equation with no internal sources 
where V and σ denote the electrical potential and the tissue
conductivity, respectively. For each of the electrode confi-
gurations, we calculated the induced E-field distribution
inside the head using the preconditioned conjugate gradient
solver within ANSYS.
E. Regional Analysis of E-Field Strength
The E-field was sampled in manually segmented ROIs
thought to be relevant to therapeutic and/or side effects of
ECT, including dorsolateral prefrontal cortex (DLPFC),
frontal pole, hippocampus, hypothalamus, insula, first dorsal
interosseous (FDI) motor area, orbitofrontal cortex (OFC),
subcallosal cingulate cortex (SCC), and thalamus. We com-
puted the median E-field strength in these ROIs in the left and
right hemispheres for the BL, BF, RUL, and FEAST elec-
Fig. 1. Diagram of the methods for generating a realistic FE head model for E-field simulation and region of interest (ROI) analysis of ECT.
Fig. 2 shows the FE head model (left), the four simulated
ECT electrode montages (BL, BF, RUL, and FEAST; upper
row), and the corresponding E-field magnitude distributions
in a representative axial slice (lower row). Fig. 2 demon-
strates that different ECT electrode configurations result in
substantially different E-field distributions in the brain.
Table I gives the median E-field strengths in the ROIs for
the various electrode configurations. The median E-field
strength in the whole brain is 3.5, 1.3, 1.7, and 2.8 V/cm, and
the right/left hemisphere median E-field ratio is 1.0, 0.9, 2.7,
and 1.8 for BL, BF, RUL, and FEAST ECT placements,
respectively. As expected, the lateralized configurations
(RUL and FEAST) manifest stronger E-field magnitudes in
the right hemisphere, whereas the symmetric configurations
(BL and BF) have comparable E-field strengths in both he-
mispheres. The frontal (DLPFC, frontal pole, and OFC) to
temporal (hippocampus) E-field ratio is 0.9, 3.7, 0.8, and 1.8
for BL, BF, RUL, and FEAST, respectively. Configurations
with anterior electrodes (BF and FEAST) induce a stronger
E-field in the frontal pole and DLPFC than configurations
with more posterior electrodes (BL and RUL). In hippo-
campus, BL produces 1.4–3.1 times stronger E-field than the
other configurations. RUL has weaker E-fields compared to
BL everywhere except in the right motor area, where it is 1.7
times stronger. Compared to the other configurations, FEAST
produces 2.3–2.7, 1.4–2.7, and 1.2–2.5 times stronger E-field
in SCC, right OFC, and right DLPFC, respectively.
IV. DISCUSSION AND CONCLUSION
A. Implications for ECT Technique
The spatial E-field distribution is a key aspect of dosage in
ECT. To the best of our knowledge, this is the first quantita-
tive study of the regional differences in E-field strength re-
sulting from variations in ECT technique.
The ECT electrode placement associated with the most
pronounced cognitive side effects, BL, produces the strongest
E-field overall and in the hippocampus. Electrode placements
associated with lower cognitive side effects (RUL and BF)
have the lowest E-fields overall and in hippocampus. BF and
FEAST have higher E-field strength in prefrontal structures
than BL and RUL, but BF achieves this with the highest
fronto-temporal ratio with maximal sparing of hippocampus.
The electrode placement with lowest seizure threshold (RUL)
has the strongest E-field in the motor strip, thought to be the
site of seizure initiation. FEAST produces more than two
times stronger E-field than any other configuration in SCC, a
target for deep brain stimulation in depression ; in right
DLPFC, a target for transcranial magnetic stimulation ;
and in right OFC, which is also implicated in network dy-
sregulation associated with depression . However, the
clinical effects of FEAST have yet to be reported.
These clinical correlations of the distinct E-field topogra-
phies associated with variation in ECT technique illustrate
how realistic E-field models can biophysically explain ob-
served differences among conventional electrode positions
(BL, BF, and RUL), as well as evaluate a novel configuration
(FEAST). Thus, the realistic head model provides quantita-
tive insight in the biophysics of ECT, which may shed light
on the differential clinical outcomes seen with various forms
of ECT. Ultimately, this work may guide the development of
Fig. 2. Left: 3-D rendering of FE head model. Right: Axial view of the ECT electrode configurations (upper row) and simulated E-field magnitude distri-
butions for BL, BF, RUL, and FEAST configurations (lower row).
SIMULATED MEDIAN E-FIELD (V/CM) FOR FOUR ECT ELECTRODE
CONFIGURATIONS AND ROIS IN LEFT (L) AND RIGHT (R) HEMISPHERES
L R L
Whole brain 3.5 3.6 1.3
DLPFC 3.5 3.7 4.9
FDI motor area 3.3 3.3 1.2
Frontal pole 4.0 4.1 10.4 11.2 1.2
Hippocampus 4.2 4.4 1.5
Hypothalamus 2.7 2.9 1.3
Insula 3.5 3.9 1.6
OFC 4.6 5.0 5.0
SCC 1.8 2.0 1.6
Thalamus 2.3 2.0 0.9
novel stimulation paradigms with improved risk/benefit ratio.
It is acknowledged, however, that these simulations address
only the E-field and not the seizure itself, the topography of
which is also thought to be a major contributor to clinical
B. Model Validation
The simulated E-field strengths are in good agreement with
previous studies using a realistic head model  and spherical
head models , . Our E-field results also find some
support in published experimental studies. The current den-
sity induced by a frontal–occipital electrode configuration
was measured in an electrolytic tank containing human
half-skull . After scaling the applied current to ECT
levels (800 mA), the measured E-field in the electrolytic tank
is in the range of 1.5–2.5 V/cm. Moreover, intracerebral
voltage measurements in human cadavers yield E-field
strength estimates of 0.7–1.8 V/cm for 800 mA of the current
applied through bifrontotemporal electrodes , . The
higher overall E-field strengths in our results (median
1.3–3.5 V/cm) are likely due to the truncation of the head
model and variations of the segmented skull thickness, which
are discussed below.
Several factors should be considered for future refinements
of this model. We modeled the electrodes as point sources
instead of pads or disk electrodes. Even though the point
source electrode approximation is not expected to substan-
tially influence the resultant E-field in the brain due to dif-
fusion of the current in the scalp and the skull, modeling a
realistic electrode shape should provide more accurate re-
sults. Another limitation is the truncation of the head model.
MRI was only acquired for the portion of the head above the
orbits; thus, the truncated head model eliminates shunting of
the ECT stimulus current in the lower portion of the head,
resulting in increased current density and E-field strength in
the brain. Further, the head model truncation has differential
effect for various ECT electrode placements. For example, in
a spherical head model analysis (not shown), we observed
that truncation of the head model results in a 13% and 56%
increase in peak E-field for RUL and BL ECT, respectively.
Finally, the E-field strength and distribution is highly sensi-
tive to skull thickness . Therefore, the precision of the
E-field simulation could be improved with more accurate
MRI segmentation  and with skull extraction from
computed tomography head scans that are co-registered with
the MRI  and which provide a strong signal from bone.
 H. A. Sackeim, "Convulsant and anticonvulsant properties of
electroconvulsive therapy: towards a focal form of brain stimulation,"
Clin. Neurosci. Res., vol. 4, no.1, pp. 39-57, 2004.
 Z.-D. Deng, A. V. Peterchev, and S. H. Lisanby, "Focality of neural
stimulation with magnetic seizure therapy (MST) and electroconvulsive
therapy (ECT) in humans and non-human primates," in 64th Ann. Conv.
Soc. Biol. Psychiatry, Vancouver, Canada, 2009.
 Z.-D. Deng, S. H. Lisanby, and A. V. Peterchev, "Effect of anatomical
variability on neural stimulation strength and focality in
electroconvulsive therapy (ECT) and magnetic seizure therapy (MST),"
in Conf. Proc. IEEE Eng. Med. Biol. Soc., 2009, pp. 682-688.
 R. N. Holdefer, R. J. Sadleir, and M. J. Russell, "Predicted current
densities in the brain during transcranial electrical stimulation," Clin.
Neurophysiol., vol. 117, no.6, pp. 1388-1397, 2006.
 M. Nadeem, T. Thorlin, O. P. Gandhi, and M. A. Persson,
"Computation of electric and magnetic stimulation in human head using
the 3-D impedance method," IEEE Trans. Biomed. Eng., vol. 50, no.7,
pp. 900-907, 2003.
 M. Sekino and S. Ueno, "Comparison of current distributions in
electroconvulsive therapy and transcranial magnetic stimulation," J.
Appl. Phys., vol. 91, no.10, pp. 8730-8732, 2002.
 W. H. Lee, T. S. Kim, A. T. Kim, and S. Y. Lee, "3-D diffusion tensor
MRI anisotropy content-adaptive finite element head mdoel generation
for bioelectromagnetic imaging," in Conf. Proc. IEEE Eng. Med. Biol.
Soc., 2008, pp. 4003-4006.
 W. H. Lee, T. S. Kim, M. H. Cho, Y. B. Ahn, and S. Y. Lee, "Methods
and evaluations of MRI content-adaptive finite element mesh
generation for bioelectromagnetic problems," Phys. Med. Biol., vol. 51,
no.23, pp. 6173-6186, 2006.
 T. F. Oostendorp, Y. A. Hengeveld, C. H. Wolters, J. Stinstra, G. van
Elswijk, and D. F. Stegeman, "Modeling transcranial DC stimulation,"
in Conf. Proc. IEEE Eng. Med. Biol. Soc., 2008, pp. 4226-4229.
 C. H. Wolters, A. Anwander, X. Tricoche, D. Weinstein, M. A. Koch,
and R. S. MacLeod, "Influence of tissue conductivity anisotropy on
EEG/MEG field and return current computation in a realistic head
model: A simulation and visualization study using high-resolution
finite element modeling," NeuroImage, vol. 30, no.3, pp. 813-826,
 P. J. Basser, J. Mattiello, and D. LeBihan, "MR diffusion tensor
spectroscopy and imaging," Biophys. J., vol. 66, no.1, pp. 259-267,
 T. Spellman, A. V. Peterchev, and S. H. Lisanby, "Focal electrically
administered seizure therapy: A novel form of ECT illustrates the roles
of current directionality, polarity, and electrode configuration in seziure
induction," Neuropsychopharm., vol. 34, no.8, pp. 2002-2010, 2009.
 H. S. Mayberg, "Targeted electrode-based modulation of neural circuits
for depression," J. Clin. Invest., vol. 119, no.4, pp. 717-725, 2009.
 J. P. O'Reardon, H. B. Solvason, P. G. Janicak, S. Sampson, K. E.
Isenberg, Z. Nahas, W. M. McDonald, D. Avery, P. B. Fitzgerald, C.
Loo, M. A. Demitrack, M. S. George, and H. A. Sackeim, "Efficacy and
safety of transcranial magnetic stimulation in the acute treatment of
major depression: A multisite randomized controlled trial," Biol.
Psychiatry, vol. 62, no.11, pp. 1208-1216, 2007.
 S. Rush and D. A. Driscoll, "Current distribution in the brain from
surface electrodes," Anesth. Analg., vol. 47, no.6, pp. 717-723, 1968.
 J. W. Smitt and C. F. Wegener, "On electric convulsive therapy with
particular regard to a pariental application of electrodes controlled by
intracerebral voltage measurements," Acta. Psychiat. Neurol., vol. 19,
pp. 529-549, 1944.
 F. M. Lorimer, M. M. Segal, and S. N. Stein, "Path of current
distribution in brain during
Electroencephalogr Clin. Neurophysiol., vol. 1, no.3, pp. 343-348,
 A. Datta, V. Bansal, J. Diaz, J. Patel, D. Reato, and M. Bikson,
"Gyri-precise head model of transcranial direct current stimulation:
Improved spatial focality using a ring electrode versus conventional
rectangular pad," Brain Stim., vol. 2, no.4, pp. 201-207, 2009.
 M. Chen and D. J. Mogul, "A structurally detailed finite element human
head model for simulation of transcranial magnetic stimulation," J.
Neurosci. Meth., vol. 179, no.1, pp. 111-120, 2009.