Exploring Brain Functional Connectivity in Hand Motion and Motor Imagery through fNIRS Signals: A Graph Theory Approach

Abstract

Using functional near-infrared spectroscopy (fNIRS) as a complementary and cost-effective neuroimaging technique in sensorimotor tasks due to its applications in brain-computer interface (BCI) research can provide useful information about functional connectivity of brain networks. However, few studies on brain functional connectivity during sensorimotor tasks have often focused on evaluating brain activity electrically. In the present study, a signal processing algorithm using fNIRS-HbO2 data has been suggested to find active parts of the brain for motion and motor imagery in motor imagery task. In this algorithm, first, the wavelet transform was used to remove the noise and preprocess the signal. Then, using correlation analysis, functional connectivity matrices in motion and motor imagery were extracted, and finally, global efficiency ​​(GE) values were calculated. In addition to investigating the conditions of the small-world network in the connectivity matrix, the classification of motion and motor imagery was investigated using a t-test. For this purpose, a 20-channel fNIRS signal was recorded to measure changes in HbO2 concentration in the motor cortex of 12 healthy individuals with a sampling frequency of 10 Hz. The results, in addition to confirming the presence of a small-world network in the graphs from the correlation matrix, showed that the classification of motion and motor imagery of right and left hands will be significant when 40% of the strongest connectivity between channels was selected. The results showed that in the left hemisphere there was stronger connectivity between the channels. In general, the results not only showed the activity of brain networks in performing sensorimotor tasks as small-world networks, but they also reported the role of the dominant hemisphere in performing these tasks.