Epilepsy Detection using Combination DWT and Convolutional Neural Networks Based on Electroencephalogram

Abstract

At the present day, smart technology has made life simpler for people in all spheres of life, including medical. It is necessary to have technology that can identify diseases or physical defects in humans since this will influence the course of therapy. One of the cutting-edge technologies used to identify epilepsy is the electroencephalogram (EEG). The signal was obtained by observed brain’s electrical activity for a period of time to get these signals. Medical professionals need to be very accurate and confident in their ability to categorize EEG patterns in order to diagnose epilepsy. This study suggested using Zero Crossing Frequency and Mean Crossing Frequency features extracted from transformed singnal using Discrete Wavelet Transform. EEG signals were classified into three categories: ictal, pre-ictal, and normal using Convolutional Neural Network. According to the study’s findings, the suggested approach can accurately categorize three categories with a confidence interval (CI) of 0.0013 and an accuracy of 98.09%.