A Robust Constrained Method for the Extraction of P300 Subcomponents

Separation of event related potentials (ERPs) is being investigated in this paper by means of a new constrained blind source extraction (BSE) technique. Specifically, the P3a and P3b subcomponents are extracted from an ERP incorporating some prior knowledge of the desired signals. The method reliably extracts the desired P300 subcomponents with great accuracy. One advantage of BSE algorithms over blind signal separation (BSS), which have been used in EEG research, is that the algorithm focuses on one source instead of trying to extract all of the sources simultaneously which involves indeterminacy in the number of sources. To testify the performance of the algorithm, some experiments are shown which are performed on real EEG data