Fuzzy clustering of EEG signal and vigilance performance

An automatic method for classification of EEG data, based upon segmentation of the signal using the autoregressive model and decision making in fuzzy environment, is described. The classification is applied to explore the relations between EEG states during waking, and vigilance performance studied through auditory choice reaction time. The average auditory choice reaction time measured during occurrences of "alpha" segments was significantly shorter than that measured during occurrences of "nonalpha" signal segments. A significant negative correlation was also found between the segments auditory choice reaction time and the segments spectral power in the alpha or beta frequency band.