Classification of brain hemodynamic signals arising from visual action observation tasks for brain–computer interfaces: A functional near-infrared spectroscopy study

• We present a real-time control of a haptic device via a new fNIRS based BCI framework.
• A visual action observation task is proposed and verified its feasibility empirically.
• A set of algorithms are employed for decoding of brain signals for external commands.

A brain–computer interface (BCI) is a way of translating an individuals’ thoughts to control a computer or an external mechanical device. Studying brain activities in a reproducible manner, this study explores the possibility of using real-time functional-near infrared spectroscopy (fNIRS) to detect brain hemodynamic features for BCI commands. Sixteen channel brain activities associated with two distinct mental tasks were measured from seven healthy subjects. The tasks represented neural activities arising from a visual observation of a motor action related to hand movements of the subjects. Sensitive signatures of task relevant neural activities were further extracted from hemodynamic signals in the prefrontal cortex of the brain, and subsequently were translated into pre-determined computer commands using a set of algorithms. The decoded commands allowed volunteer subjects to control an external device in real-time through their mental intentions. The obtained results demonstrate the potential of the current study as an alternative fNIRS-BCI paradigm.