Epileptic Automated Detection from EEG Signal Using Statistical Features and Machine Learning Technique

Epileptic seizures detection based on EEG signals is crucial in the diagnosis of patients with epilepsy. Experts mainly employed the visual inspection to identify epileptic seizures; however, it is a tedious job for them. This study, proposed designing an automated model to detect if the patient has epilepsy or not this could support the clinical research, facilitates the task of experts and speeds up the detection process. Thus, the aim of this research is to develop an automatic model to detect epileptic seizures from EEG signals. In this study, we present a new method for seizure classification for EEG signals using a statistical features coupled with a least square support vector machine (LS_SVM) classifier. To achieve this task, each single EEG channel has been divided into four clusters. In addition, each cluster is segmented into sub-intervals. A vector of statistical features is pulled out from each sub-cluster form the final features set. The obtained features set are sent to the LS-SVM classifier. In this paper, epileptic database from Bonn University is used to evaluate the performance of the proposed model. In this paper, our developed model has been tested with epileptic dataset and the proposed model produced an average accuracy, sensitivity and specificity of 100%, 100% and 100%, respectively.