Functional near-infrared spectroscopy based brain activity classification for development of a brain-computer interface

Research in brain-computer interface (BCI) has increased recently. It allows the user to communicate with a computer/external device through the process of thinking. Functional near-infrared spectroscopy (fNIRS) is a relatively new non-invasive optical imaging modality that can be used to measure cortical brain activities. The main advantages of using this technique are relatively low cost, safety, portability and wearability. In this paper we propose to apply fNIRS to measure different brain activities during thinking task and body motion task. For thinking experiment the human subjects were asked to solve simple arithmetic calculations. For the body motion task, the subjects were asked to tap finger after regular intervals of time. Continuous-wave imaging system (DYNOT: dynamic near-infrared optical tomography) was used to image each subject´s prefrontal cortex during the arithmetic task and motor cortex during the finger tapping task. Our results of fNIRS signal analysis showed different patterns of hemodynamic response for the different tasks that can be used to recognize the brain tasks. Using Fisher´s linear discriminant analysis we were able to distinguish clearly between the different brain activities with an average accuracy of above 80%. This idea can be used to directly control a robot through thinking process or through movement of the body using fNIRS.