International Congress on Telecommunication and Application’14 University of A.MIRA Bejaia, Algeria, 23-24 APRIL 2014 Temporal Epilepsy Seizure Prediction Using Graph and Chaos Theory T. Haddad, L. Talbi, A. Lakhssassi Université du Québec en Outaouais Département d’informatique et d’ingénierie 101, St-Jean-Bosco, Gatineau, PQ, J8Y 3G5 Canada N. Ben-Hamida, S. Aouini Ciena Canada 3500 Carling Ave. Ottawa, ON, K2H 8E7 Canada Abstract— Temporal seizures due to Hippocampal origins are very common amongst epileptic patients. This article presents a novel seizure prediction approach based on a combination of graph and Chaos theory. The early identification of seizure signature allows for various preventive measures to be undertaken. The proposed non-linear approach consists of observing a high correlation level between any pair of electrodes for the lower frequencies and a decrease in the Lyapunov index (chaos or entropy) for the higher frequencies. The Fast Fourier Transform (FFT) and statistical analysis tools were used to determine threshold levels for the lower frequencies. A graph topology characterizes seizure signatures for each patient. These graphs are used to monitor both Delta and Gamma behaviors. In order to validate the proposed approach, six patients from both sexes and various age groups with temporal epilepsies originating from the hippocampal area were studied using the Freiburg Database. An average seizure prediction of 30 minutes, a detection accuracy of 72%, and a false positive rate of 0% were accomplished throughout 200 hours of recording time. Keywords: Graph Theory; Epilepsy Prediction; EEG Signal; Chaos theory. I. INTRODUCTION Epilepsy is a neurological dysfunction originating from a systemic nervous disorder causing recurrent seizures to its victims. Seizure triggering can be due to either physiological or environmental causes and their repetition can occur in several week intervals or even several hours. Loss of conscience and equilibrium happening during seizures can lead to serious injuries, fractures or even burns. Health costs linked to these accidents are very important. In this paper, a novel procedure for anticipating seizures in temporal epilepsy is described. The methodology uses both Graph and Chaos theory to predict upcoming seizures. The article is divided as follows: Section 2 highlights the state-of- the-art in seizure prediction. Section 3 briefly explains Signal Processing for Seizure Prediction along with a brief overview of graph theory. Section 4 details our anticipation methodology. Section 5 presents the experimental results using EEG recordings from the Freiburg Database (http://epilepsy.uni-freiburg.de/) [1]. Results are analyzed and a discussion comparing the proposed technique with similar works is presented. II. STATE-OF-THE-ART AND BACKGROUND EEG is the summation of neuronal electrical activities and is widely used in diagnosing epileptic disorders and seizure onsets. During a seizure, brain waves differ considerably from the normal state. Indeed, differences in amplitude, phase, correlation, spectral density and chaos rate are observable due to the abnormal neuronal firing. Brain activity in the ictal, interictal and healthy states are significantly different (Fig. 1). Fig. 1: EEG spectral density during interictal phase (Red), pre-ictal phase (Bleu) and postictal phase (Green) III. Neural synchronization in a network is important for the normal functioning, especially for the information processing. It can also reflect abnormalities due to epilepsy. Zhou et al. and Zemanova et al. [2, 3] studied correlations