ITERATIVE MUSIC FOR HIGHLY CORRELATED EEG/MEG SOURCE LOCALIZATION
This study presented an iterative MUSIC (Multiple Signal Classification) for highly correlated EEG source localization. By suppressing the equivalent false source, the approximate true source location information was obtained. And then, by iteratively suppressing source found in the last iteration, eventually, both of the sources were identified. The method is designed to tackle highly correlated sources, for example, bilateral activations at primary auditory/auditory cortices, at which cases conventional MUSIC has difficulty. Compared with other similar methods, the presented one needs less computation load since it utilizes the minor difference between sources, as can be adequately explained by a theoretical model for correlated sources. Simulation and real data test confirmed its effectiveness.