A graph representation of functional diversity of brain regions

Introduction: Modern network science techniques are popularly used to characterize the functional organization of the brain. A major challenge in network neuroscience is to understand how functional characteristics and topological architecture are related in the brain. Previous task-based functional neuroimaging studies have uncovered a core set of brain regions (e.g., frontal and parietal) supporting diverse cognitive tasks. However, the graph representation of functional diversity of brain regions remains to be understood. Methods: Here, we present a novel graph measure, the neighbor dispersion index, to test the hypothesis that the functional diversity of a brain region is embodied by the topological dissimilarity of its immediate neighbors in the large-scale functional brain network. Results: We consistently identified in two independent and publicly accessible resting-state functional magnetic resonance imaging datasets that brain regions in the frontoparietal and salience networks showed higher neighbor dispersion index, whereas those in the visual, auditory, and sensorimotor networks showed lower neighbor dispersion index. Moreover, we observed that human fluid intelligence was associated with the neighbor dispersion index of dorsolateral prefrontal cortex, while no such association for the other metrics commonly used for characterizing network hubs was noticed even with an uncorrected p <.05. Conclusions: This newly developed graph theoretical method offers fresh insight into the topological organization of functional brain networks and also sheds light on individual differences in human intelligence.