Detection of Epileptic Seizures from EEG Signals Using EM Algorithm and Frequency Analysis

This paper proposes a new method based on Expectation Maximization algorithm (EM) and Gaussian Mixture model for classification of electroencephalogram (EEG) signals. The detection of epileptic form discharges in the EEG is an important component in the diagnosis of epilepsy. Decision making was performed in two stages: feature extraction using the Fast Fourier Transform (FFT) first and then the probability density functions values are trained with the Expectation Maximization algorithm as a clustering method. Two types of EEG signals were used as input for the classifier with two discrete outputs: normal and epileptic. At last the performance of the proposed method is proved in terms of classification accuracies by simulation results. The results confirmed the maximum mean classification accuracy of 100% in classifying of the EEG signals.

Keywords:

Electroencephalogram (EEG) , Fast Fourier Transform (FFT) , Expectation-Maximization (EM) algorithm , Epileptic seizure