Automatic detection of epileptic sharp-slow by wavelet and approximate entropy

The automatic epileptiform activities detection in EEG is significant in clinical application. Epileptic sharp-slow complex wave is one of typical presence of epileptiform activities, which has different time-frequency property compared with spike and spike-slow complex wave. A new scheme is presented for detecting epileptic sharp-slow wave in 8-channel EEG data from normal subjects and epileptic patients. The scheme is based on the characteristic of a multi-resolution and approximate entropy (ApEn) analysis of EEG signals. The EEG signals on each channel are decomposed into three levels using multi-resolution wavelet analysis, and then ApEn values of the detail coefficients are computed. Distinct differences are found between the ApEn values of the epileptic sharp-slow and the normal EEG. The EEG signals are detected by Neyman-Pearson (NP) criteria. The optimal detection rule of detecting sharp-slow is achieved, and it assures a higher detection rate with a lower false detection rate.