DETECTION OF EPILEPTIC EVENTS IN ELECTROENCEPHALOGRAMS USING WAVELET ANALYSIS

This study deals with the problem of identification of epileptic event s in electroencephalograms using multiresolution wavelet analysis. The following problems are analyzed: time localization and characterizati on of epileptiform events, and computational efficiency of the method. The algorithm presented is based on a polynomial spline wavelet trans form. The multiresolution representation obtained from this wavelet tr ansform and the corresponding digital filters derived allows time loca lization of epileptiform activity. The proposed detector is based on t he multiresolution energy function. Electroencephalogram records from epileptic patients were analyzed, and results obtained are shown. Some comparisons with other methods are given.