Nonlinear Analysis of EEG for Detection of Seizure Onset Zone

Abstract

A complex system like brain is expected to be nonlinear and it incorporates the non stationary nature of EEG signal. EEG time series signal shows very small changes during non seizure and seizure states. Marking these subtle changes is difficult in linear methods. Fractal dimension (FD) is a measure of such complexity and intermittent transitions in time domain. Hence nonlinear method of FD computation has been used for the identification of seizure onset zone. Results show that the method can determine pre seizure, seizure, post seizure and non seizure phases. The main part of the paper presents the detection of the seizure onset zone using this method. The detection of EEG seizure is verified with the gold standard values given in the data sets and with specialist’s expertise. Seizure detection and seizure prediction are challenging task in EEG signal analysis. Identification of EEG signals particularly early seizure form the basis for understanding mechanism leading to seizure condition. In this proposed research work Children's Hospital Boston and the Massachusetts Institute of Technology (CHB-MIT) database are used.