부산대학교 도서관

Brain-computer interface (BCI) is a well-established technology that facilitates the communication between a user and an external device solely based on brain activity, bridging users' intentions from a variety of human brain signals, including EEG (Electroencephalogram), fNIRS (functional near-infrared spectroscopy), and DTI (diffusion tensor imaging). Out of these, EEG, a technique to record electrical brain activities using a noninvasive electrophysiological method that measures voltage fluctuations induced by the ionic current within brain neurons, is the most commonly applied method. With no clinical risk, EEG data can be recorded using affordable acquisition equipment and is highly portable. Among the various paradigms of EEG, Motor Imagery (MI) has garnered a lot of recognition in the last ten years. Owing to its potential, several ground-breaking research transforming human life have been conducted resulting in world-class BCI products. In this paper, we provide a comprehensive overview of the various EEG-based MI-BCI classification trends and challenges with a particular emphasis on deep learning approaches.