Epileptic Seizure Classification Using Crisscross Strategy-based Gannet Optimization Algorithm with Temporal Convolutional Network

Epilepsy seizure is a serious neurological condition that refers to abnormal activities in the brain cells which are identified by ElectroEncephaloGram (EEG) signals for the automatic classification of brain seizures.However, traditional seizure classification methods fail to classify brain seizure effectively due to the variations in EEG data for different patients.In order to overcome this limitation, a Crisscross Strategy-based Gannet Optimization Algorithm with Temporal Convolutional Network (CGOA-TCN) is proposed for effective feature selection and epileptic seizures classification using EEG signals.The d GOA algorithm selects diverse features based on various duration recordings of EEG data using the crisscross strategy.Followed by that, the TCN model efficiently classifies the defective and non-defective seizure signals with its ability to learn long-range dependencies that enhance the model's classification performance.The EEG signals are preprocessed using a Butterworth filter to remove noise and other irrelevant artifacts.Then, a Discrete Wavelet Transform (DWT) is used for feature extraction from different domains, which are then selected and classified by employing the proposed model.Experimental results of the CGOA-TCN prove that the model attains an accuracy of 98.96 % and 99.89% on the CHB-MIT and BONN datasets, respectively, outperforming existing approaches like Hierarchical Long Short Term Memory (Hierarchical LSTM).