Expert system design for classification of brain waves and epileptic-seizure detection

Feature extraction and classification of electro-physiological signals is an important issue in development of disease diagnostic expert system (DDES). Classification of electroencephalogram (EEGs) signals (normal and abnormal) is still a challenge for engineers and scientists. Various signal processing techniques have already been proposed to solve this puzzle of classification of non linear signals like EEG. In this work, attempts have been taken to distinguish between normal, epileptic and non-epileptic EEG waves by use of Support Vector Machine (SVM). EEG signals from (healthy subject with eye open condition, healthy subject with eye close condition, signal from hippocampus region and signal from opposite to epileptogenic region and signal with seizure) were considered for the analysis. The signals were processed by using wavelet-chaos techniques. The nonlinear dynamics of the original EEGs are quantified in the form of the correlation dimension (CD, representing system complexity) and the largest Lyapunov exponent (LLE, representing system chaoticity), Capacitive Dimension (CAD) which show the randomness nature of the signal. SVM classifier applied on the extracted feature vectors for the classification purpose. From the results, it was clearly found that the classification accuracy was significantly higher i.e. more than ninety percentage. Hence the techniques can be implemented to design knowledge based expert disease diagnostic system.