Detection of epileptic seizure in EEG signals using window width optimized S-transform and artificial neural networks

The occurrence of epileptic seizure in EEG segment is a nonstationary process. As the nonstationary EEG signal contains multiple frequencies, the conventional frequency based methods cannot be used for their analysis. In this paper, we propose window width optimized S-transform based method for epileptic seizure detection and compare its performance with standard S-transform and Short time Fourier transform (STFT) based detection methods. The detection of epileptic seizure is performed in five stages - (i) Optimization of S-transform, (ii) Time-frequency representation of EEG segments using window width optimization of S-transform, (iii) Calculation of Power spectrum density (PSD), (iv) Feature extraction, and (v) Classification of seizure containing EEG segment using Artificial Neural Network (ANN). The performance of proposed method has been evaluated for three classification problems along with a comparison with other time-frequency methods.