Least Square Support Vector Machine for Interictal Detection Based on EEG of Epilepsy Patients at Airlangga University Hospital Surabaya-Indonesia

Epilepsy is a chronic disease characterized by recurrent seizures. Epileptic seizures occur due to central nervous system (neurological) disorders. Around 50 million people worldwide suffer from epilepsy. The diagnosis of epilepsy can be done through an electroencephalogram (EEG). There are two important periods to consider in EEG recording, the interictal period (clinically no seizures) and ictal (clinically seizures). Meanwhile, visual inspection of EEG signals to detect interictal and ictal periods often involves an element of subjectivity and it requires experience. So that automatic detection of interictal periods with classification method is badly needed. In this study, Least Square Support Vector Machine (LS SVM) method for classification of interictal and ictal was used. Data preprocessing process was carried out using Discrete Wavelet Transform (DWT). The result showed that the classification using LS SVM with kernel RBF method achieved an accuracy under curve (AUC) of 96.3%.