Prognosis of epileptic seizures using EEG signals

Electroencephalogram is a pathological phenomena that has chaotic properties and characterized as a nearly random signal as the neurons are theoretically highly non linear. The non-linear analysis method is effectively applied to electroencephalogram signals to study the dynamics of the complex underlying behavior. Analyzing EEG signals with the aid of nonlinear dynamics takes the advantage of requiring much lower quantity of data. The objective of our research is to analyze the acquired EEG signals for non linearity by applying signal processing tools and use the non linear features for design of CAD system to classify epileptic seizures. To ensure the soundness of the proposed method and the suitability of the selected parameters, exhaustive statistical analysis of the descriptor components was performed. Highest classification accuracy of 98.4% was obtained with MLPNN classifier with high sensitivity and specificity followed by SVM and KNN classifiers. The promising performances observed are demonstrative of the efficiency and efficacy of systems developed for classification and prediction of normal and ictal conditions of epileptic patients.