Epileptic Seizure Detection using EEG Signals based on 1D-CNN Approach

Abstract

Epilepsy is one of the chronic neural conditions branded by an excessive and uncontrolled electrical explosion in the brain; it appears as seizures. This anomaly affects almost <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> percent of the world. For this reason, seizure detection has become a subject of interest in the last decade, to perform these analyzes, study characteristics of brain activity, furthermore neurological disorders, and especially epileptic seizures electroencephalography (EEG) is used. Diverse scientific methods have been used to reliably detect epileptic seizures due to EEG Signals. In this analysis, deep learning based on CNN 1D convolutional neural networks were developed, as used as models for DL, VGGNET, and ResNet. The tests were conducted using standard data sets. The proposed method was exercised on 23 subjects of the CHBMIT dataset, which successfully achieved an average accuracy of 97.60 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> and 97.32% respectively for ResNet and VGGNET. The results obtained suggest the effectiveness of the use of ResNet in epileptic seizure detection.