Comparison of simulated electrode configurations
for transcranial stimulation

P. Kauppinen¹, P. Laarne², M. Tenhunen², S. Oja^3
1Ragnar Granit Institute, Tampere University of Technology,
P.O. Box 692, Fin-33101 Tampere, FINLAND
²Department of Clinical Physiology, ³Department of Neurology, Tampere University Hospital,
P.O. Box 2000, Fin-33521 Tampere, FINLAND

Abstract: Somatosensory evoked potential and motor evoked response monitoring are important methods in diminishing the risk of neurologic injury during surgical procedures. We have applied a realistic, 3-D head model based on finite difference method to study various electrode configurations generally used for stimulation. Our initial results are in line with clinical experience. Future studies are called to obtain optimal stimulation electrode configurations for various stimulation purposes.

INTRODUCTION

During scoliosis surgery it is important to monitor the state of spinal cord. The lack of oxygen can cause unrecoverable changes in nerves and damage the spinal cord. Use of somatosensory evoked potential and motor evoked response monitoring is an important part of intraoperative routine. Effects of various stimulation waveform or multiple waveforms and their amplitude are increasingly being studied[1], but with few electrode configurations.

The depolarization of cortical motor neurons depends, in addition to the form of the pulse, on achieved stimulation current density in the region of interest in motor cortex. In order to achieve efficient stimulation with minimal current, selective electrode configurations should be used for stimulation. So far, electrode configurations used for stimulation have mainly been based on intuitive understanding on generated current distribution inside head or on experimenting and observing obtained responses with various electrode configurations.

To allow reliable spinal cord monitoring, it is important that the stimulation setup used during surgery is optimal. We have conducted a preliminary comparison study on various electrode configurations used for the transcranial electrical stimulation by applying a three-dimensional anatomically accurate computer head model in solving the stimulation fields associated with each configuration.

METHODS

Computer Conductivity Model for the Head

A realistic, 3-D head model based on finite difference method (FDM) was constructed from whole head magnetic resonance image set having 112 transversal slices [2]. The electrode locations of 10-10 electrode system were marked by oil filled capsules prior to imaging session (Fig. 1a).

The scalp, skull, cerebrospinal fluid (CSF), gray matter, white matter and the anatomical cavities were classified using IARD segmentation method [3]. Resolution of model was downscaled for computational purposes by combining four pixels to one. This final model consisted of about 7 mm³ voxels resulting to 435369 nodes for the FDM calculations.

Figure 1.  a) Appearance of the FDM model with the 10-10 EEG electrode locations, b) Electrode locations used in simulating current fields, reference in all simulations was Cz.

Simulation of Existing Electrode Configurations

Applying the FDM [4], current fields were simulated for each electrode location shown in Fig. 1b). Figure 2 shows studied electrode configurations derived from the International 10-20 electrode system often used in stimulation studies. Since our model is linear, these configurations can be derived from the set of original simulated current fields.

Figure 2.  Studied stimulation electrode configurations, marked from EC1 through EC6. Reference in each case is marked with black spot.

Analysis

Simulated current fields were investigated in the region of interest, which was divided into 11 segments (Fig. 3). Relative currents were calculated in each segment. In addition to total current, currents were analyzed independently in x, y and z-directions. Actual amount of current administered to these regions was not investigated, since only comparison of studied configurations was done. Also, to express homogeneity of current delivery, standard deviations (STD) for obtained current values were calculated.

Figure 3.  Approximate illustration of the region of interest, divided into 11 subvolumes, superimposed on an MR image of the subject.

RESULTS

Main results of the simulation study are given in Table I. Highest total current was obtained with the EC5, the value being approximately twice the one obtained with the EC4. Lowest STD with second highest overall current delivery was obtained with the EC6, indicating high current delivery in all directions in the region of interest, investigated as a whole.

TABLE I
Simulated current distributions of studied electrode configurations. Given values are averages
in all segments and normalized, independently for I_total and I in each direction.

Config.	Total current	IX	IY	IZ	STD
EC1	0.94	0.54	0.29	1.36	0.47
EC2	0.84	0.45	0.59	1.16	0.31
EC3	0.88	0.49	0.40	1.26	0.40
EC4	0.68	3.24	0.26	0.12	1.46
EC5	1.47	0.48	3.19	0.57	1.26
EC6	1.20	0.80	1.27	1.52	0.30

DISCUSSION

The data derived from the present study demonstrate the feasibility of modeling in analyzing stimulation electrode configurations. Differences in the current strength within the studied configurations were not especially large, indicating the anticipated result from the experimental knowledge. All studied configurations are being used clinically, since there is not a single configuration outperforming others. On the other hand, Ubags et al. [1] demonstrated that configuration EC6 increased stimulation response as compared to a two-electrode configuration. Our results support their finding, since EC6 produced second largest total current and, large currents simultaneously in each direction. Looking at e.g. a segment deepest in the brain, EC6 produced highest stimulation current. Bipolar configurations EC4 and EC5 produced highest total current strengths in the direction of the measurement, which is intuitively also reasonable.

Results obtained are only preliminary, with several limitations. First of all, the resistivity values used in simulations are only rough approximations of real values, and the resolution of the model might not be accurate enough to include for instance the effects of well-conducting cerebrospinal fluid in the simulations. Orientation of neural tissue in the region of interest has not been considered, either, which may be important in repolarization process. Most importantly, the region of interest was considered as a whole, not as a function of distance from the cortical surface.

Nevertheless, not a single electrode configuration produced exceptional current strengths. Furthermore, the results indicate that computer modeling could be applied in searching optimal stimulation electrode configuration. An algorithm could be implemented, that combines the electrodes of the 10-10 system in various ways and then analyses the corresponding stimulation capabilities. Most selective configurations could then be investigated in clinical environment for their usefulness in practice.

Acknowledgments: Work was supported financially by the Ragnar Granit Foundation, Finnish Cultural Foundation and the Medical Research Fund of Tampere University Hospital.

REFERENCES

[1]  L.H. Ubags, C.J. Kalkman, H.D. Been, J.C. Drummond, "The use of a circumferential cathode improves amplitude of intraoperative electrical transcranial myogenic motor evoked responses," Neurosurgical Anesthesia, vol. 82, pp. 1011-1014, 1996.

[2]  P.H. Laarne, M.L. Tenhunen-Eskelinen, J.K. Hyttinen, H.J. Eskola, "Effect of EEG electrode density on dipole localization accuracy using two realistically shaped skull resistivity models," Brain Topography, vol. 12, 249-254, 2000.

[3]  T. Heinonen, P. Dastidar, P. Kauppinen, J. Malmivuo, H. Eskola, "Semi-automatic tool for segmentation and volumetric analysis of medical images," Medical & Biological Engineering & Computing, vol. 36, pp. 291-296, 1998.

[4]  P. Kauppinen, J. Hyttinen, P. Laarne, J. Malmivuo, "A software implementation for detailed volume conductor modelling in electrophysiology using finite difference method," Computer Methods and Programs Biomedicine, vol. 58, pp. 191-203, 1999.