Detection Of Normal And Epileptic EEG Signals Using By Lifting Based DWT Transform And Neural Network

Abstract

Electroencephalograms are electric measurements of brain waves, commonly used in
diagnosis of epilepsy. We investigate soft computing techniques for the rapid classification of
epilepsy risk levels from EEG signals to evaluate them. This paper presents a three-class
classification system based on discrete wavelet transform (DWT) and Neural Network. Researchers
have recently suggested several studies on feature extraction, feature selection and classification
strategies for epileptic seizure detection. Features for epilepsy detection are derived from EEG in
the time-domain, frequency-domain and wavelet domain. It may not be relevant or possible for
classification, considering all the features extracted from EEG, as it is time consuming. Proper
selection of characteristics therefore takes on significance. We extracted mean, variance, Entropy
and Standard deviation from the EEG and fed them to the classifier. The following stages are
proposed 1. Data collection, 2.Feature extraction, 3.Classification. The purpose of the paper is to
include an automated device to assist a doctor in the process of diagnosis.