Classification of prefrontal and motor cortex signals for three-class fNIRS–BCI

• Feasibility of three-class fNIRS–BCI demonstrated.
• Classification of fNIRS signals corresponding to three different brain activities.
• Motor and prefrontal cortex activities used.
• Intentionally-generated cognitive tasks as inputs.

Functional near-infrared spectroscopy (fNIRS) is an optical imaging method that can be used for a brain-computer interface (BCI). In the present study, we concurrently measure and discriminate fNIRS signals evoked by three different mental activities, that is, mental arithmetic (MA), right-hand motor imagery (RI), and left-hand motor imagery (LI). Ten healthy subjects were asked to perform the MA, RI, and LI during a 10 s task period. Using a continuous-wave NIRS system, signals were acquired concurrently from the prefrontal and the primary motor cortices. Multiclass linear discriminant analysis was utilized to classify MA vs. RI vs. LI with an average classification accuracy of 75.6% across the ten subjects, for a 2–7 s time window during the a 10 s task period. These results demonstrate the feasibility of implementing a three-class fNIRS-BCI using three different intentionally-generated cognitive tasks as inputs.