Evaluating the Accuracy of an Anisotropic Finite-Volume Head Model for the EEG Forward Problem

Realistic head models are required to accurately solve the electroencephalogram (EEG) forward problem. We have developed a finite volume method (FVM) head model for this purpose. It uses a high-resolution millimeter-level cubic grid, incorporates anisotropic conductivities, and is both physically intuitive and simple to implement. In this work we use an analytic spherical head model to evaluate the performance of our FVM formulation. Two different error measures related to the magnitude and shape of the scalp potentials are used to compare the numerical and analytic head models. We determine that our FVM head model agrees with the analytic head model and calculates the shape of the scalp potentials more accurately than the magnitude. As well, our head model performs better at higher resolutions and severe anisotropy appears to be a problem at the current resolutions.