Uncovering Community Structure in Neuronal Functional Networks from Multi-neuronal Spike Trains

Abstract

Brain functional network is an important method to study the structure of brain function and brain diseases. In this paper, we propose a new neuronal functional network community structure detection method. First, we use the random walk distance to determine similarity matrix. Finally, we use the spectral decomposition method to decompose the new similarity matrix. We automatically determine the number and structure of neuronal functional networks. We evaluate the performance of this method on surrogate datasets, which know the structure in advance. We apply this method on the recorded spike trains in rat. The rat performed the Y-maze behavioral task. We find the community structures from neuronal functional networks. The traditional clustering methods cannot find these structures.