Automatic detection of photic evoked spikes contaminated with slow burst by use of morphological filter and similarity coefficient

A method for detecting spikes and slow burst in photic evoked electroencephalogram (EEG) was proposed. The spikes were detected by combining methods of the morphological filter and the similarity coefficient in the time domain. The slow burst was detected by using pole of AR model (auto-regressive model) in the frequency domain. The proposed method was applied for the photic evoked EEG data containing spikes and slow burst, and brought satisfactory coincidence with the results interpreted by a qualified electroencephalographer.