Finger-Induced Motor Imagery Classification from Hemodynamic Response Using Type-2 Fuzzy Sets

摘要

Recent advances in brain-computer interface (BCI) and advanced computational algorithms make us capable of classifying different human motor imageries based on their corresponding noninvasive brain signals. Although there exists significant research results on left-/right-hand motor imageries, there is a scarcity of research on the classification of individual finger movements by motor imagery techniques. This paper provides a solution to this open problem by proposing an advanced classifier to classify the finger motor imageries corresponding to the motor intentions at individual fingers of the right hand with the help of functional near-infrared spectroscopy (fNIRS)-based hemodynamic response of the human brain. Experimental results obtained confirm that the proposed hemodynamic response-based classification outperforms the reported results of electroencephalography (EEG)-based classification in terms of classification accuracy. Statistical tests included confirm the efficacy of the proposed technique over its competitors.