Estimating cortical connectivity in functional near infrared spectroscopy using multivariate autoregressive modeling

Performing cognitive tasks requires close interaction between cortical regions of the brain. Monitoring the functional connectivity during a particular task could contribute to a better understanding of the role and interactions of underlying brain regions. In this paper, we employ time variant Multi-variate Autoregressive (MAR) modeling to establish functional connectivity between regions involved in a cognitive task. Data is collected from neonates whose brain activity was monitored by functional Near Infrared Spectroscopy (fNIRS) while being exposed to 2 different types of auditory stimuli. The method was applied to data from 3 subjects on a predefined set of channels known to be involved in auditory and structural processing. The connectivity analysis reveals a common connectivity pattern among the subjects which is neuroanatomically and functionally relevant. Moreover, investigation of temporal evolution of connectivity between temporal and frontal areas shows an increase in connection strength towards the end of the experiment.