A signal processing based analysis and prediction of seizure onset in patients with epilepsy

// Hamidreza Namazi 1 , Vladimir V. Kulish 1 , Jamal Hussaini 2, * , Jalal Hussaini 3, * , Ali Delaviz 4, * , Fatemeh Delaviz 5, * , Shaghayegh Habibi 6, * , Sara Ramezanpoor 6, * 1 School of Mechanical and Aerospace Engineering, Nanyang Technological University, Singapore 2 Faculty of Medicine, Universiti Teknologi MARA (UiTM), Selangor, Malaysia 3 Department of Otolaryngology, Faculty of Medicine, Shiraz University, Shiraz, Iran 4 Faculty of Mechanical Engineering, Iran University of Science and Technology, Tehran, Iran 5 Faculty of Chemical Engineering, Islamic Azad University (Fars Science and Research Branch), Shiraz, Iran 6 School of Medicine, Fasa University of Medical Sciences, Fasa, Iran * These authors have contributed equally to this work Correspondence to: Hamidreza Namazi, e-mail: m080012@e.ntu.edu.sg Keywords: epileptic seizure, prediction, EEG signals, the Hurst exponent, fractal dimension Received: October 16, 2015 Accepted: November 09, 2015 Published: November 17, 2015 ABSTRACT One of the main areas of behavioural neuroscience is forecasting the human behaviour. Epilepsy is a central nervous system disorder in which nerve cell activity in the brain becomes disrupted, causing seizures or periods of unusual behaviour, sensations and sometimes loss of consciousness. An estimated 5% of the world population has epileptic seizure but there is not any method to cure it. More than 30% of people with epilepsy cannot control seizure. Epileptic seizure prediction, refers to forecasting the occurrence of epileptic seizures, is one of the most important but challenging problems in biomedical sciences, across the world. In this research we propose a new methodology which is based on studying the EEG signals using two measures, the Hurst exponent and fractal dimension. In order to validate the proposed method, it is applied to epileptic EEG signals of patients by computing the Hurst exponent and fractal dimension, and then the results are validated versus the reference data. The results of these analyses show that we are able to forecast the onset of a seizure on average of 25.76 seconds before the time of occurrence.