A method for extracting subspace of deterministic sources from EEG data

In this paper, an algorithm for separating linear subspaces of time-locked brain responses and other noise sources in multichannel electroencephalography data is proposed. The search criterion used by method discriminates time-locked brain components and noise components on the basis of the assumed deterministic behavior that the time-locked brain sources obey. The comprehensive derivation of the method is given together with the description and the analysis of the results of the method´s application to simulated and real EEG data sets. The possibilities of improving the results are also discussed.