Comprehensive Study of Brain–Computer Interface (BCI) with Emphasis on Strategies Used for Processing the Signals

Abstract

Brain–computer interfaces (BCI) can be considered as a developing field, as it is evolving on almost all fronts, including types of input signals and the number of control signals derived from them, the method for acquiring the signals, signal processing methodologies used, etc. This chapter deals with electroencephalogram-based (EEG) non-invasive BCI system; keeping this consideration in mind, an exhaustive literature survey is carried out to throw light on different types of BCI systems, different available brain signals that can be used as inputs, various signal processing methodologies used for signal pre-processing, or signal modeling and feature extraction. Brain signals that can be potentially utilized for BCI, such as evoked potential and spontaneous signals, are explored in the chapter. P300 evoked potential, steady-state visual evoked potential (SSVEP), and slow cortical potential (SCP) are explained as evoked potential and sensorimotor rhythms (SMR) as spontaneous signals. Signals are judged based on the task accuracy achieved in classifying them and the transfer of information obtained by use of these signals in the system, referred to as information transfer rate (ITR). Various promising algorithms for signal modeling, extracting features, and classification are sequentially explored. Time, frequency, and spatial domain methods are elaborated and research gaps in the methods used are specified; thus, helping in selection of the method for implementation or modifications can be suggested. This study follows the guideline given by Mason and Birch regarding signal processing requirements. The study of classification techniques included in the chapter used for EEG-based BCI covers a wide range of linear and non-linear classifiers with supervised and unsupervised learning. This study provides an outline that will help in comparing the system being test as well as helping develop the methodology selection for a new system.