Tracking progression of schizophrenia using a resting-state fMRI biomarker of regional interactions in the brain network.

Schizophrenia is a chronic mental disorder thought to affect cognitive processes and emotional regulation by disrupting communication between brain regions. The high level of training and experience required for accurate diagnosis limits access to care for many patients with this debilitating illness, leading to delays in diagnosis and progression of illness for an uncertain period. To improve accuracy of treatment, we investigated a potential quantitative method of tracking progression of schizophrenia using resting-state fMRI. Using data sourced from the DecNef rsfMRI open dataset in High and Low Duration cohorts, we constructed personalized dynamic network models that characterize influence between cortical regions of the brain. The contrasting levels of influence were converted to a phase space and ranked according to a novel network-based biomarker we call the "sink index." When the sink index is high it suggests that a region is being heavily influenced by other parts of the network and is not itself influencing the network strongly. Out of seventy cortical regions, the sink index of the left banks of the superior temporal sulcus was identified as able to significantly differentiate between cohorts and built a classifier of very high accuracy (sens 0.86, spec 1.0, AUC 0.99). Our results support the hypothesis that the pathophysiology of schizophrenia is indicative of aberrant network connectivity patterns.