Performance Enhancement of a Brain-Computer Interface using High-Density Multi-Distance NIRS

This study investigated the effectiveness of using high-density multi-distance source-detector (SD) separations in near-infrared spectroscopy (NIRS) for enhancing the performance of functional NIRS (fNIRS)-based brain-computer interface (BCI). The NIRS system that was used for the experiment is capable of measuring signals from four SD separations: 15, 21.2, 30 and 33.5 mm; this allowed the measurement of hemodynamic response alterations at various depths. Fifteen participants were asked to perform mental arithmetic and word chain tasks to induce task-related hemodynamic response variations or asked to stay relaxed to acquire a baseline signal. To evaluate the degree of BCI performance enhancement by the high-density channel configuration, the classification accuracy obtained using the typical low-density lattice SD arrangement was compared with that obtained using the high-density SD arrangement, while maintaining the SD separation at 30 mm. The analysis results demonstrated that the use of high-density channel configuration did not result in noticeable enhancement of classification accuracy; however, combining hemodynamic variations measured by two multi-distance SD separations resulted in significant enhancement of the overall classification accuracy. The results of this study indicate that the use of high-density multi-distance SD separations is likely to provide a new method of enhancing the performance of an fNIRS-BCI.