A Comparative Study of Feature Extraction Methods in P300 Detection

In this paper some different feature extraction methods are compared and their performances in a patternrecognition based P300 detection system are studied. By studying the features in different domains it was concluded that time domain features are more powerful in discriminating P300 signals from non-P300 signals. Therefore, three different sets of features were considered in the time domain and the performance of each was assessed by Fisher’s linear discriminant (FLD) classifier, the best set being identified based on this assessment. The experiment was also performed in two phases each with a different number of channels to analyze the effect of the number of channels on performance.

Keywords:

ERP , Brain Computer Interface (BCI) , P300 Detection , Feature Extraction