Convolutional Neural Network in Brain-computer Interfaces-exoskeleton System

Abstract

Brain-computer interfaces (BCIs) have emerged as a groundbreaking technology that has the potential to revolutionize the field of stroke rehabilitation. These innovative systems allow individuals who have suffered from strokes to regain lost motor function by directly connecting their brains with external devices, such as exoskeletons. One of the most commonly used paradigms in BCIs is motor imagery (MI), which involves generating electroencephalograms (EEGs) through imagined movements. This means that stroke patients can perform motor tasks simply with the help of exoskeletons by only thinking about them, without any physical movement required. The ability to decode these EEG signals is crucial for enabling effective communication between the brain and external devices. Deep learning (DL), particularly convolutional neural networks (CNNs) which are able to extract meaningful features from raw EEG data and accurately classify different types of imagined movements, has proven to be highly effective in decoding EEG signals generated during motor imagery tasks and has already made numerous contributions in this area. This article provides an initial overview of BCIs and CNNs, followed by an explanation of how CNNs decode EEG signals derived from motor imagery, as well as the utilization of BCIs for controlling exoskeleton devices. Finally, the current limitations and future directions in this field are discussed.