Using Local Minimum and Maximum Points in EEG for Diagnosis of Epilepsy

Abstract

Epilepsy is a neurological disorder, disrupting nerve cell activity and causing seizures and sometimes loss of consciousness, affecting 0.6-0.8 % of the world of population. The most common method for the diagnosis of epilepsy is electroencephalography (EEG). With the help of EEG, nerve cell activity can be monitored noninvasively in a practical way. EEG signals are low amplitude bioelectric signals detected by electrodes from the brain surface. The amplitude of these signals is 1-400 μV from peak to peak, and the frequency band is in the range of 0.5-100 Hz. In this study, it is aimed to diagnose epilepsy from EEG signals obtained from seizure-free periods of people diagnosed with epilepsy. The peaks and troughs of the EEG signals obtained from seizure-free periods of healthy and epileptic individuals have been used as feature. The peaks and troughs, which are determined as feature, have been trained in the Artificial Neural Network (ANN) with the Levenberg-Marquardt model. Achieving 90% accuracy at the end of the training has shown that local maximum and local minimum points can be used to diagnose epilepsy from EEG signals.