EEG Classification of Epileptic Patients Based on Signal Morphology

Abstract

Epilepsy is a chronic neurological disorder, which potentiates the occurrence of seizures in its victims. This clinical picture significantly affects the daily lives of patients, which in some cases, the occurrence of seizures can promote injury, trauma or even sudden death. People who suffer from epilepsy could live free of seizures, if they were previously diagnosed or given appropriate treatment. Therefore methodologies that simplify and expedite the diagnosis and treatment of these individuals are valid and necessary. This study aims to develop a classification model of electroencephalograms (EEG) with presence or absence of epileptic seizures. It was adapted a methodology of extraction of characteristics used for classification of electrocardiograms (ECG) with or without cardiac arrhythmias, which is based on the calculation of statistical moments in time windows. After defining the statistical characteristic that promotes greater separation between the two groups, this characteristic was calculated from a set of 119 electroencephalograms from the CHB-MIT database. The acquired characteristic vectors were classified using linear classifiers, reaching 97% accuracy. The results suggest that the proposed classification methodology can be used to aid in the diagnosis of patients with suspected epilepsy.