Automatic epileptic signal classification using deep convolutional neural network

Epilepsy is a neurological illness that causes seizures in the brain and affects a huge number of people worldwide. Electroencephalography (LEG) is the most commonly used modality for epilepsy prognosis, although visual inspection of EEG signals is a time-consuming and cumbersome task. To avoid that, several automated systems have been developed to assist neurologists. Feature extraction-based machine learning algorithms were used long before the advent of deep learning. But their success was limited to the capabilities of those who crafted the features manually. Deep learning is an artificial intelligence branch in which feature extraction and classification are completely automated. This paper, in particular, presents a deep learning architecture, Convolutional Neural Network (CNN), to classify EEG signals into three categories: normal, pre-ictal, and ictal or seizure. The proposed model achieved an accuracy, precision, recall, F-measure, and error rate of 94.0%, 93.2%, 94.3%, 93.7, and 6.0% respectively.