ABSTRACT

Localization of cortical sources in scalp electroencephalography (EEG) is of fundamental importance in EEG research. The EEG source localization problem has two major components: forward problem and inverse problem. The purpose of the forward problem is to make a model of the head of the subject whose EEG is being collected. A more realistic head model will be the one in which the skull is subdivided into smaller pieces, each having a different conductivity. If the pieces are small enough, the conductivity may reasonably be taken to be uniform across that small piece. Source localization by minimum norm inverse can be further improved with additional constraints. For example, it is possible to improve the accuracy of source localization by phase synchronization and signal power. Multiple signal classification is used to describe techniques for localization of sources of multiple waveforms arriving at an array of sensors.