A brain-computer interface (BCI) is a system that records brain activity, processes it, and identifies commands that are delivered to external devices or applications to carry out the desired action of a user. BCI has the potential to restore the lost functions of disabled people, improving the users’ quality of life. EEG-based BCI systems use electroencephalographic (EEG) signals to identify the user’s intention, which requires applying multiple computational methods to process the EEG electrical activity. Currently, important limitations remain in using EEG signals for BCI applications, including the number of mental states that can be classified, the interference of other signals (artifacts) that contaminate the EEG, the percentage of effectiveness in generating commands, and the low rate of information transfer. Processing EEG signals include stages for trial/artifact rejection, epoching, filtering, feature extraction, and classification/regression. This chapter describes EEG signal processing pipelines used in the scientific literature and presents new approaches for EEG data analysis using emerging computational methods applied to BCI applications focused on restoring motor functions. Deep learning methods and multimodal signal fusion for BCI will be discussed.
