Modularity maps reveal community structure in the resting human brain

Abstract

The brain is a complex network of interconnecting neurons that combines regional specificity with distributed processing. Recent advances in the field of network theory have facilitated ground-breaking analyses demonstrating that brain connectivity exhibits small-world properties similar to other self-organized networks such as the internet, the genome, or even social organizations. Brain connectivity supports local and global processing through high clustering and short connectivity paths, respectively. While these comprehensive network indices highlight the global organization of the network, the regional specificity is related to the interconnectivity of local neighborhoods within the global system. The work presented here evaluated the community structure of resting human brain networks to identify the local neighborhoods and map those interconnected areas back to the brain. The study identified predictable clustering in unisensory cortices. However, the unexpected community structure in the default-mode network (DMN) revealed three separate modules and included the lateral frontal cortices in addition to traditional DMN regions. These results are the first to map modularity across the entire brain without restricting analyses to predefined anatomical structures. Such analyses provide an unbiased view of network communities and promise to provide new insights into organization of the brain. Evaluation of modular brain structure across states, during demanding tasks, or in disease populations will reveal dynamic connectivity changes in whole-brain networks.