Background: Brain network dynamics are responsive to task induced fluctuations, but such responsivity may not hold in schizophrenia (SCZ). We introduce and implement Centrality Dynamics (CD), a method developed specifically to capture task-driven dynamic changes in graph theoretic measures of centrality. We applied CD to functional MRI (fMRI) data in SCZ and Healthy Controls (HC) acquired during associative learning.
Methods: fMRI (3T Siemens Verio) was acquired in 88 participants (49 SCZ). Time series were extracted from 246 functionally defined cerebral nodes. We applied a dynamic windowing technique to estimate 280 partially overlapping connectomes (with 30 135 unique region-pairs per connectome). In each connectome, we calculated every node's Betweenness Centrality (BC) following which we built 246 unique time series from a node's BC in successive connectomes (where each such time series represents a node's CD). Next, in each group similarities in CD were used to cluster nodes.
Results: Clustering revealed fewer sub-networks in SCZ, and these sub-networks were formed by nodes with greater functional heterogeneity. The averaged CD of nodes in these sub-networks also showed greater Approximate Entropy (ApEn) (indicating greater stochasticity) but lower amplitude variability (suggesting less adaptability to task-induced dynamics). Finally, higher ApEn was associated with worse clinical symptoms and poorer task performance.
Limitations: Centrality Dynamics is a new method for network discovery in health and schizophrenia. Further extensions to other task-driven and resting data in other psychiatric conditions will provide fuller understanding of its promise.
Conclusion: The brain's functional connectome under task-driven conditions is not static. Characterizing these task-driven dynamics will provide new insight on the dysconnection syndrome that is schizophrenia. Centrality Dynamics provides novel characterization of task-induced changes in the brain's connectome and shows that in the schizophrenia brain, learning-evoked sub-network dynamics were (a) less responsive to learning evoked changes and (b) showed greater stochasticity.
Keywords: clinical psychiatry; functional MRI; neuroscience; schizophrenia; statistical methods.