Semi-automatic epilepsy spike detection from EEG signal using Genetic Algorithm and Wavelet transform

Abstract

A novel algorithm is proposed for identifying epileptic features in electroencephalograph (EEG) signals automatically. The proposed algorithm is based on the combination of the Genetic Algorithm (GA) and the Wavelet transform. Optimal Wavelet basis functions that adapt the spikes of the EEG signal are first designed using GA. Then they are used as matched filters to identify the spikes related to seizure activity from the EEG recordings using Wavelet transform and a threshold-based estimation method. The method can estimate the number and the location of epileptic spikes in an EEG signal very fast and almost in real time. Hence, it is suitable for data mining of EEG recordings of epileptic patients for fundamental studies of epilepsy, prediction of seizures, and treatment of epilepsy. We have applied and evaluated the method using different samples of real clinical EEG data of epileptic patients, where it has shown a very high sensitivity (more than 90%) and selectivity (more than 90%).