Deep learning for EEG Channel Selection for Epilepsy Detection and Classification

Abstract

Recent and past researches have shown an increasing number of patients who are affected by epilepsy or epileptic seizure, which is a neurological disorder. Electroencephalogram (EEG) is a well-known technique or procedure that is effective, non-invasive, and widely used in many studies for detection and classification of Epilepsy. Continuous recording of EEG data provides opportunity for further analysis to better manage epilepsy by both the clinicians and patient’s family. As per the present clinical practice, neurologists spend significant time to analyze each channel manually to identify the presence of the epileptic events in the hours of the recording. This challenge has motivated the development of automatic detection systems for epilepsy detection and classification. Though there are existing many approaches for epilepsy detection and classification, with the recent history still it is a problem due to its large amount of non-stationary data. Recently, Deep Learning (DL) has shown very high accuracy for the image classification and time series analysis. DL has also been used for many other Brain Signal Processing applications giving promising results. So, this paper considers the review of DL applications for Epilepsy detection and Classification to understand the scope, improvements and limitations. This will help the research community to identify the research gaps and future research directions.