Linear Analysis of Epileptic Forms in the Electroencephalogram

Abstract

In our study, we investigated the properties of electroencephalogram (EEG) of humans experiencing epileptic seizures. By applying linear methods, such as spectral analysis and wavelet transformation, we compared characteristics of EEGs of patients suffering from epilepsia with the characteristics of EEGs of healthy individuals. Our goal was to establish the real effect of epileptic attacks in frequency domains, distinguishing them from various types of artifacts caused by mental and physical activities in humans. The frequency range of 1 to 2 Hz was found to be interesting, where we have detected a power spectral density peak typical of epileptic seizures. Also, we have shown that there exists an anomalous distribution of the relative wavelet energy of the signal within this frequency range. We also showed that the frequency range 7.97 to 15.94 Hz can serve as an indicator of preictal activity in the EEG signal.