Feature extraction methods for electroencephalography based brain-computer interface: A review

Introduction: A brain-computer interface (BCI) is a rapidly growing cutting-edge technology in which a communication pathway is built between the human brain and computer. The BCI is also known as a direct neural interface where user can control external devices with the help of the brain signals. Neural signals are typically measured using electroencephalography (EEG). Objective: Feature extraction from EEG data performs a significant role in the wearable BCI computing field. Since a large amount of EEG data, the major challenge is the effective feature extraction and reduce the computation burden. The objective of this paper is to review such different feature extraction techniques for the development of effective and robust BCI systems. Approach: We reviewed feature extraction techniques employed in EEG based BCI studies. We synthesize these studies in order to present the taxonomy and report their usage with pros and cons. Significance: This paper provides a comprehensive review of feature extraction techniques for EEG based BCI with their properties. Furthermore, open challenges are also discussed for further advancement in BCI studies. © International Association of Engineers.