Clustering-led complex brain networks approach.

Abstract

This paper reviewed the meaning of the statistic index and the properties of the complex network models and their physiological explanation. By analyzing existing problems and construction strategies, this paper attempted to construct complex brain networks from a different point of view: that of clustering first and constructing the brain network second. A clustering-guided (or led) construction strategy towards complex brain networks was proposed. The research focused on the discussion of the task-induced brain network. To discover different networks in a single run, a combined-clusters method was applied. Afterwards, a complex local brain network was formed with a complex network method on voxels. In a real test dataset, it was found that the network had small-world characteristics and had no significant scale-free properties. Meanwhile, some key bridge nodes and their characteristics were identified in the local network by calculating the betweenness centrality.