Improving classification accuracy of covert yes/no response decoding using support vector machines: An fNIRS study

One of the aims of brain-computer interface (BCI) is to restore the means of communication for people suffering severe motor impairment, anarthria, or persisting in a vegetative state. Yes/no decoding with the help of an imaging technology such as functional near-infrared spectroscopy (fNIRS) can make this goal a reality. fNIRS is a relatively new non-invasive optical imaging modality offering the advantages of low cost, safety, portability and ease of use. Recently, an fNIRS based online covert yes/no decision decoding framework was presented [Naseer and Hong (2013) online binary decision decoding using functional near infrared spectroscopy for development of a braincomputer interface]. Herein we propose a method to improve support vector machine classification accuracies for decoding covert yes/no responses by using signal slope values of oxygenated and deoxygenated hemoglobin as features calculated for a confined temporal window within the total task period.