An Xception Spiking Fractional Neural Network-Based Framework for Accurate Detection and Classification of Epileptic Seizures in Electroencephalogram Signals

Abstract

Epileptic Seizures in Electroencephalogram Signals (EEG) is a neurological disorder that can be detected by continuous observation of the brain signals using EEG, allowing for early treatment and personalized care. The conventional approach for seizure detection, including visual analysis and manual analysis of EEG signals, offers useful information but is prone to limitations like noise in signals, inter individual variability, and the need for correct diagnosis during seizures. To address these challenges, an xception spiking fractional neural network-driven framework for accurate detection and classification of epileptic seizures in electroencephalogram signals (DCES-EEG-XSFNN) is proposed. At first, input data is gathered from Bangalore EEG Epilepsy Dataset. These data are pre-processed utilizing Robust Maximum Correntropy Kalman Filter (RMCKF) which is used for data normalization. Then the pre-processed data are given to Convolutional Variational Attention Transformer (ConVAT) for feature extraction, which is employed to extract relevant features. The extracted features are fed to Xception Spiking Fractional Neural Network (XSFNN) for detection and classification, which is used to detect epileptic seizures and classify such as seizure events, focal seizures, generalized seizures and healthy. To further improve the performance of XSFNN during detection and classification, the Superb Fairy-wren Optimization Algorithm (SFOA) is utilized. The proposed technique implemented in python, demonstrates substantial improvements in accuracy, precision, recall, F1-Score and Confusion matrix. The proposed achieves the best results accuracy of 97.67%, precision of 99.34%, recall of 97.67%, and an F1-score of 97.5% for Seizure compare with existing methods such as Enhanced EEG Signal Processing for Accurate Epileptic Seizure Detection (EEG-ESD-SVM), Epileptic Seizure Detection in EEG Signals Using Machine Learning and Deep Learning Techniques (ESD-EEG-1DCNN), and Epileptic Seizure Detection in EEG Signals Using Machine Learning and Deep Learning Techniques (DES-EEG-LSTM).