Iterative joint dynamic brain mapping and neural activity modeling from electroencephalographic signals

Abstract

A novel joint iterative dynamic inverse problem solution for brain mapping based on electroencephalographic (EEG) signals is presented. Proposed approach considers linear and nonlinear time-varying state space models of the brain as dynamic constraints in the solution of the dynamic inverse problem where the brain mapping and the neural activity model are estimated simultaneously from EEG signals. The method performance is evaluated in terms of standard error, projection error, and residual error for several SNRs by using simulated EEG signals. As a result, a considerable improvement over Low Resolution Tomography (LORETA) and Dynamic LORETA approaches is found.