Performance analysis of ApEn as a feature extraction technique and time delay neural networks, multi layer perceptron as post classifiers for the classification of epilepsy risk levels from EEG signals

Abstract

Epilepsy being a very common and chronic neurological disorder has a pathetic effect on the lives of human beings. The seizures in epilepsy are due to the unexpected and transient electrical disturbances in the cortical regions of the brain. Analysis of the Electroencephalography (EEG) Signals helps to understand the detection of epilepsy risk levels in a better perspective. This paper deals with the Approximate Entropy (ApEn) as a Feature Extraction Technique followed by the possible usage of Time Delay Neural Network (TDNN) and Multi Layer Perceptron (MLP) as post classifiers for the classification of epilepsy risk levels from EEG signals. The analysis is done in terms of bench mark parameters such as Performance Index (PI), Quality Values (QV), Sensitivity, Specificity, Time Delay and Accuracy.