非表示:
キーワード:
-
要旨:
Brain networks are characterized by strong recurrence, and widespread connectivity. As a consequence it is inherently difficult to tell apart local processing and interactions between structures. This is a major obstacle to the identification of a modular organization of the brain. However, complex network analysis enables to attack the problem from a different angle. Specifically, such analysis may consider directly the whole brain as a network and then characterize its topology.
In this work, we use this framework to identify the polysynaptic topology of functional brain networks with a high spatial resolution. We first estimated network connectivity from fMRI signals by computing statistical dependency measures between pairs of voxels. Then, assuming that a restricted set of core regions relay information to the whole network, we developed a statistical test to characterize the structure of this high dimensional network using the concept of eigenvector centrality [1].
We applied these techniques to fMRI recordings in 6 humans during resting state and 4 monkeys during anesthesia. Eigenvector centrality measures based on correlation enabled us to identify a robust set of central areas that was similar in both species, involving cortical (precuneus, medial prefrontal cortex) and subcortical structures (hippocampus). Further graph theoretic analysis based on random walks allowed clustering these regions into robust groups with dedicated subnetworks of influence and to identify their hierarchical organization (clusters of central regions in human are shown in the figure below).
In sum, centrality revealed a synthesis of the complex topology of functional networks in a consistent restricted set of core regions in monkey and human brains. Further work will investigate the temporal dynamics of these regions, and their influence on the activity of the whole network will be validated by experimental manipulation.
