Oppositional Crow Search Algorithm Based Artificial Neural Network for Epileptic Seizure Classification

The Electroencephalogram (EEG) is a widely used diagnostic test for identifying epilepsy by analyzing the electrical activity in a person's brain. Epilepsy poses a significant social challenge due to its unpredictable nature, high treatment costs, and the urgent need for improved diagnostic methods. In this context, a novel approach is proposed namely, Oppositional Crow Search Algorithm (OCSA)-based Artificial Neural Network (ANN) for classifying epileptic seizure disorders. Data is gathered from the CHB-MIT dataset, subjected to signal pre-processing, and enhanced through Empirical Mode Decomposition (EMD) for increased training data volume. Features are extracted, including entropy, Hjorth parameters, fractal dimension, as well as Hurst exponent. Subsequently, the proposed OCSA-ANN method is applied to classify epileptic seizures through a training and testing process. The results demonstrate that the OCSA-ANN approach outperforms existing methods, achieving a remarkable sensitivity of 96.71%, an accuracy rate of 89.88%, and a specificity of 89.88%.