Multi-Resolution EEG AND EEG Sub-Band Features Optimization for Epileptic Classification Using Hybrid Evolutionary Computing Technique

Abstract

Epilepsy is one of the neurological issues causing interminable irregular electrical release in the mind. The electroencephalogram (EEG) has developed as a critical instrument in the cerebrum movement checking for epilepsy conclusion. Hereditary Algorithms (GA) and Neural Networks (NN) have a place with developmental processing that attempt to estimating the neurological issue, for example, epilepsy. In this paper EEG signals broke down for multi resolution sub-band by wavelet disintegration. Wavelet procedure is executed for investigation of EEG and delta, theta, alpha, beta, and gamma sub-groups of EEG. Different features like AR coefficients, control range thickness, entropy and factual features extraction alongside morphological features. The system received to break down three distinctive dataset of EEG signals: 1) Healthy Person; 2) Epileptic Patients amid a without seizure period (interictal or pre-seizure);3) Epileptic Patients amid seizure event (ictal). After feature extraction objective is to enhance the precision of classification. Received GA for feature enhancement alongside multi-layer backpropogation ANN classifier by assessed preparing execution and characterization correctness’s and results reasoned that proposed diagnostic delicate instruments has successfully grouping EEG signals.