Localization of brain electrical activity by a combination of beamforming and independent component analysis

This paper proposed a method for brain electric activity localization by a combination of beamforming techniques and independent component analysis (ICA). By preprocessing the EEG data using ICA, the EEG data is decomposed into several fixed scalp maps and the corresponding one dimensional time courses. The product of each scalp map and the corresponding time courses generates the scalp EEG recordings responsible for only one source. And then, the conventional adaptive beamformer is repeatedly used to the generated EEG recordings to localize each topographic map due to each source. The process transforms a problem that find several local maxima within 3 dimensional brain volume into another problem that find out several global maxima in the same space. The head model simulation experiments confirm the theoretical analysis. The simulation also show that the proposed method can identify the independent source with higher spatial resolution.