Item

Abstract

A Brain-computer Interface (BCI) can be developed using the optical response of Near Infrared Spectroscopy (NIRS) which measures metabolic brain activation. NIRS can localize brain regions with a spatial resolution in mm and temporal resolution in hundreds of ms. NIRS has the advantages of noninvasiveness, portability and affordability. The authors have developed a multi-channel NIRS-BCI that distinguishes the brain activation during imagination of left hand and right hand movement. This mechanism is being incorporated in a word speller, in an on-going research project, to help severely disabled persons to communicate. Experiments with a volunteer have shown that the HMM (Hidden Markov Model) classifier performers better (average accuracy of 91.29%) than the SVM (Support Vector Machine) classifier (average accuracy of 75.62%). This might be due to considerable variations in the temporal domain in the performance of the tasks, and such variations may be better dealt with by dynamic machines like HMM. NIRS avoids the noise prominent in the EEG, and is less cumbersome to use, as there is no need for applying conducting gel. The most important advantage of the NIRS is its ability to localize brain activity noninvasively. This provides us with an excellent opportunity to use a variety of motor and cognitive tasks, and to detect signals from specific regions of the cortex for the development of powerful and user-friendly BCIs.