Spectral Entropy for Epileptic Seizures Detection

The electroencephalogram (EEG) is the brain signal that represented the valuable information about the brains condition. The configuration of the signals waveform may contain valuable and useful information about the different states of the brain. Since the biological signals are personal, indications may occur highly random in both time and frequency domains. Thus the computer analyzing is necessary. EEG is decomposed by wavelet transform and coefficient sets are obtained. In this paper spectral entropy is applied to these coefficient sets for epileptic seizures detection. This process is applied to three different groups of EEG signals: 1) healthy states, 2) epileptic states during a seizure-free interval (interictal EEG), 3) epileptic states during a seizure (ictal EEG). At the end the statistical analysis is applied for distinguishing the coefficient sets. This statistical process can differentiate between ictal and healthy subject (with eyes close) of cD2 coefficients (15-30 Hz) with 99% p-value.