Automatic Detection of EEG as Biomarker using Deep Learning: A review

Abstract

Automated Electroencephalography (EEG) is a dynamic signal that can take years of training to interpret correctly, as well as advanced signal processing and feature extraction methodologies. Due to its ability to learn good feature representations from raw data, deep learning (DL) has recently shown great promise in helping make sense of EEG signals. The question of whether DL genuinely provides advantages over more conventional EEG processing methods remains unanswered. In this paper, we examine 154 papers that use deep learning to analyses EEG data, with applications ranging from epilepsy to sleep to brain–computer interfaces to cognitive and affective monitoring. In order to inform future research and formulate recommendations, we extracted trends and highlighted DL approaches from this broad body of literature.