Physical principles of brain–computer interfaces and their applications for rehabilitation, robotics and control of human brain states

Abstract

Brain–computer interfaces (BCIs) development is closely related to physics. In this paper, we review the physical principles of BCIs, and underlying novel approaches for registration, analysis, and control of brain activity. We analyze recent advances in BCI studies focusing on their applications for (i) controlling the movement of robots and exoskeletons, (ii) revealing and preventing brain pathologies, (iii) assessing and controlling psychophysiological states, and (iv) monitoring and controlling normal and pathological cognitive activity. We consider the BCI as a hardware/software communication system that allows interaction of humans or animals with their surroundings without the involvement of peripheral nerves and muscles, using control signals generated from brain cerebral activity. Classifying BCIs into three main types (active, reactive and passive), we describe their functional models and neuroimaging methods, as well as novel techniques for signal enhancement and artifact recognition and avoidance, to improve BCI performance in real time. We also review different BCI applications, including communications, external device control, movement control, neuroprostheses, and assessment of human psychophysiological states. Then, we describe the most common techniques for the analysis and classification of electroencephalographic (EEG) and magnetoencephalographic (MEG) data. Special attention is paid to modern technology based on machine learning and reservoir computing. We discuss main results on the creation and application of BCIs based on invasive and noninvasive EEG recordings. First, we consider neurointerfaces for controlling the movement of robots and exoskeletons. Second, we describe BCIs for diagnosis and control of pathological brain activity, in particular, epilepsy. We also discuss the results on the development of invasive BCIs for predicting and mitigating absence epileptic seizures. After that, we focus on passive neurointerfaces for assessing and controlling a person's psychophysiological states and cognitive activity. Special attention is given to optogenetic brain interfaces using photostimulation to deliver intervention to specific cell types. We outline the basic principles of optogenetic neurocontrol and extracellular electrophysiology recording. We also describe the state-of-the-art of miniaturized closed-loop optogenetic devices to control normal and pathological brain activities. Further, we discuss the new emerging technological trend in the BCI development which consists in using neurointerfaces to improve the interaction between people, so-called brain-to-brain interfaces (BBIs). Such interfaces can increase the efficiency of collaborative processes when working in a group. We propose a BBI which distributes a cognitive load among all team members working on a common task. This BBI allows sharing the workload among the participants according to their current cognitive performance, estimated from their electrical brain activity. The novel results of the brain-to-brain interaction are promising for the development of a new generation of communication systems based on the neurophysiological brain activity of interacting persons, where the BBI estimates physical conditions of each partner and adapts the assigned task accordingly. Finally, we trace the main historical epochs in BCI development and applications and highlight possible future directions for this research area, including hybrid BCIs.