Finding community of brain networks based on artificial bee colony with uniform design

Abstract

The brain networks can offer fundamental insights into healthy human cognition and the alteration in disease. Some neural unit modules in brain networks can provide us a great deal of useful information. It is appealing how to find these neural unit modules and how to partition the brain network into several dense modules. There are as high within-module densities as possible and as low between-module densities as possible in these dense modules. The modularity metrics can well evaluate whether a community is good or not. Therefore, we present a novel method to find community modules of brain networks in this study. It integrates uniform design into artificial bee colony (abbreviated as UABC) in order to maximize the modularity metrics. The difference between UABC and the other existing methods lies in that UABC is presented at the first time for detecting community modules. Several brain networks generated from functional MRI for studying Autism are used to test the proposed algorithm. Experimental results performing on these brain networks demonstrate that the proposed algorithm UABC can acquire better modularity and higher stability than other competing methods.