A DWT-BASED MULTICLASS CLASSIFICATION OF EPILEPTIC ACTIVITY IN PATIENTS

Abstract

The identification of seizure activities in non-stationary electroencephalography (EEG) is a challenging task. The seizure detection by human inspection of EEG signals is prone to errors, inaccurate as well as time-consuming. Several attempts have been made to develop automatic systems so as to assist neurophysiologists in identifying epileptic seizures accurately. The proposed study suggests using Discrete Wavelet Transform to decompose the EEG signals into frequency sub-bands. We choose a certain subset of the frequency sub-bands for feature selection. Following the DWT decomposition, we calculate the Standard Deviation for each sub-band present in the subset. Finally, the standard deviation values of the sub-bands are fed to a Support Vector Machine. The proposed work consists of 5 experiments which are essentially classification problems: 3 of which are Multi-class classification problems and the rest two are Binary Classification problems. In the proposed work, we investigate the three-class classification problems focused on classifying an EEG signal into one of the three classes, which are 1. Healthy patient 2. Seizure-free epochs:inter-ictal stage 3. Epileptic Activity:ictal stage. The dataset used in the proposed work is obtained from the Department of Epileptology of the University of Bonn.The accuracy achieved in one of the Multi-class classification experiment in the proposed work is 98.45% which beats the state of the art accuracy in this three-class problem. Additionally, the proposed method has achieved highest accuracy of 100% in classifying normal EEG signals(eyes closed) and seizure EEG signal and an accuracy of 100% in classifying normal EEG signals(eyes open) and seizure EEG signal which is comparable with the existing state of the art EEG signal classification techniques. Six different classification techniques have been used in each of the five experiments conducted where every classification technique has been used with 8 different Daubechies wavelets db1 to db8. The results obtained from these experiments provide valuable insights establishing that SVM performs the best in most of the experiments with the db4 wavelet among the 8 wavelets achieving the highest accuracy.