Harnessing Physiological Responses to Improve NIRS-Based Brain-Computer Interface Performance

Abstract—This study investigated the inclusion of physiological responses due to mental activity tasks as complementary in-formation for improved-performance brain-computer interfaces (BCI) based on near-infrared spectroscopy (NIRS). Cortical hemodynamic responses due to music imagery tasks, combined with physiological responses, were collected from six able-bodied participants. Hidden Markov models (HMM) were used to model the normative physiological and hemodynamic responses of the individual at rest. A normalized log-likelihood measured was proposed for automated detection of music imagery events, which in turn, were used as BCI control signals. Experimental results suggest that improved performance is attained once physiological signals are incorporated into the BCI design. More specifically, improvements of 13 % and 7 % were observed in system sensitivity and specificity, respectively. These results suggest that physiologi-cal signals can be used to enhance the performance of NIRS-based BCIs and thus provide a more reliable communicative channel for individuals with severe motor impairments. I