T-weighted Approach for Neural Information Processing in P300 based Brain-Computer Interface

A novel method for feature extraction based on T-statistic criterion is put forward and introduced for P300 potential detection in brain-computer interface (BCI) applications. After decorrelation by principal component analysis (PCA), the optimized weighted sum of EEG signal is computed to construct the features. Applied to P300 speller paradigm of BCI competition 2003 and BCI competition III (2005), this method achieved character accuracy of 100% and 90% respectively, and by the latter score our group got the third place for the P300 dataset (dataset II) in the BCI competition III