Modeling brain electrical activity by an image-based boundary element method

Electroencephalography (EEG) source localization of brain activity is of high diagnostic value. This work aims to improve the low-spatial-resolution scalp EEG measurement through noninvasive numerical procedures. An image-based boundary element method (BEM) is developed to reconstruct cortical brain potential distribution from the scalp input. The developed BEM circumvents a practical challenge of linking scan images to model-based computation by translating the scan surface tessellation directly into mesh discretization for the BEM. Related issues, such as the numerical regularization of the ill-posed inverse problem, which are crucial to achieving reliable solutions, are discussed. The numerical studies show that the developed BEM can effectively handle the potential reconstruction on a detailed brain surface from blurry scalp potential input, and may become a promising tool to aid clinical diagnosis of brain-related problems.