Epilepsy Seizure Classification Using Artificial Neural Network and Linear Discriminant Analysis Algorithm

Abstract

Epilepsy is a condition that affects 8 out of every 1000 persons on the planet. It's a condition that alters the brain's nerve cell activity, resulting in seizures in the patient. For more than three decades, people have been fascinated by the idea of detecting and forecasting epileptic episodes. According to studies, irregular brain activity occurs a few minutes before the onset of a seizure, which is referred to as the preictal stage. Electroencephalography (EEG) is an electrophysiological monitoring technique that records the electrical activity of the brain in order to detect and forecast epileptic episodes. Predicting epileptic seizures before they happen can help prevent them and guarantee adequate seizure control. Many researchers have attempted to anticipate the preictal stage of a seizure, but successful prediction with high sensitivity and specificity remains a difficulty. This research presents a machine learning model for classifying EEG signals into seizure and non-seizure data that uses Linear Discriminant Analysis (LDA) for feature extraction and an Artificial Neural Network for classification. An EEG recording of ten patients was used to test the proposed approach. The data was filtered, and features were chosen using Linear Discriminant Analysis. The data was divided into seizure and non-seizure categories using an artificial neural network. With a classification time of 0.013s to 1s, the model had an overall accuracy, sensitivity, precision, specificity, and F1-score of 86 percent, 69 percent, 80 percent, 96.2 percent, and 72 percent, respectively. The contribution of this research is the introduction of LDA for smart and