Neurological patients with severe physical impairment require a means of expressing their intentions. Using a case study, we aimed to examine the discriminative performance of brain states based on optical topography, with the prefrontal cortex as the region of interest. Brain activity in three neurological patients with residual ability to communicate was measured using near-infrared spectroscopy in the resting state and the state of performing mental arithmetic. Statistical-based features were extracted from the acquired physiological signals, and brain states were classified using machine-learning methods. The number of correct predictions by the discriminant model was 30 out of 36 for participant A (
p < .01), 18 out of 24 for participant B (
p < .05), and 48 out of 60 for participant C (
p < .01), all of which were significantly different from the results of the exact binomial test with chance probability. These results suggest that the brain-computer interface can be applied to communication based on brain activity by defining a correspondence between brain state and intention.
View full abstract