Combining the extremities on the basis of separation: a new approach to EEG/ERP source localization

Abstract Current methods for the localization of EEG and event-related potentials (ERP) sources assume that sources are either discrete (dipole-like) or distributed. While both types of sources are likely to contribute significantly to EEG and ERP signals, each method adopts only one of these models and thus may localize the sources of other type incorrectly or not find them at all. Recently introduced Independent Component Analysis (ICA) and more general approach, Blind Source Separation (BSS), make possible the separation of signals from various brain and extra-brain (related to artifacts) sources and can be used as preprocessing technique before applying the localizing algorithms. We suggest using this preprocessing step for combining different localization methods. A brain source, if extracted correctly, can be analyzed separately from the other sources, and thus, the most appropriate localization technique can be chosen for each source. Distributed sources are likely to be localized more precisely without detailed separation but after BSS “cleaning” data from strong localized sources.