EEG Signal Classification using Discrete Wavelet Transform (DWT) and Gaussian Support Vector Machine (SVM) for Epileptics

As a potentially life-threatening neurological disorder, epilepsy needs early treatment to prevent the deterioration of the patient’s condition. Electroencephalogram (EEG) technology-based detection is a recommended method for accurate diagnosis to help distinguish the state of normal patients from patients with epilepsy. In this study, EEG recording data was obtained from a database of Klinik fiir Epileptologie, University of Bonn. The data consisted of five conditions involving nono-epilepsy volunteers and patients with epilepsy engaged in different activities. Our proposed design consists of several stages, such as dataset input, decomposition, and feature extraction (including minimum, maximum, mean, standard deviation, kurtosis, and skewness), performed through Discrete Wavelet Transform (DWT) on wavelet coefficients and classified using Support Vector Machine (SVM). Our design involved the DWT utilizing Daubechies and Gaussian wavelet types, while SVM employed a classification approach. The system’s performance was evaluated using K-Fold cross-validation and Confusion Matrix analysis. The best accuracy results, obtained from normal patients and epileptics (Cl: ABCDE) with DWT and SVM methods, showed a maximum accuracy of 78.6%. This research demonstrates the effectiveness of the proposed approach in distinguishing between normal patients and those with epilepsy, indicating its significance in epilepsy diagnosis and treatment.