Time-frequency image based features for classification of epileptic seizures from EEG signals

Abstract

In this paper, a new method for the automatic classification of seizures based on time-frequency image (TFI) of electroencephalogram (EEG) signals is proposed. Automatic classification of seizures is a crucial part of the diagnosis and treatment of epileptic seizures. TFI has been obtained by the smoothed pseudo Wigner-Ville distribution (SPWVD) based time-frequency representation (TFR) of EEG signal. The contrast stretching based pre-processing is used to increase the dynamic range of image pixels of TFI. The TFI has been segmented into rhythms of EEG signals based on frequency-bands. The features extracted from segmented images have been used as input, set to least squares support vector machines (LS-SVM) classifier together with the linear kernel, radial basis function (RBF), Mexican hat wavelet, and Morlet wavelet kernel functions for automatic classification of seizure from EEG signals. The proposed method for classification of EEG signals has provided better classification accuracy than other existing methods. The experimental results are presented to show the effectiveness of the proposed method for the classification of seizure from EEG signals.