Identifying the brain's connector hubs at the voxel level using functional connectivity overlap ratio

• Main Authors: Epifanio Bagarinao, Hirohisa Watanabe, Satoshi Maesawa, Daisuke Mori, Kazuhiro Hara, Kazuya Kawabata, Reiko Ohdake, Michihito Masuda, Aya Ogura, Toshiyasu Kato, Shuji Koyama, Masahisa Katsuno, Toshihiko Wakabayashi, Masafumi Kuzuya, Minoru Hoshiyama, Haruo Isoda, Shinji Naganawa, Norio Ozaki, Gen Sobue
• Format: Article
• Language: English
• Published: Elsevier 2020-11-01
• Series: NeuroImage
• Online Access: http://www.sciencedirect.com/science/article/pii/S1053811920307278

Neuroimaging studies have shown that the brain is functionally organized into several large-scale brain networks. Within these networks are regions that are widely connected to several other regions within and/or outside the network. Regions that connect to several other networks, known as connector hubs, are believed to be crucial for information transfer and between-network communication within the brain. To identify regions with high between-network connectivity at the voxel level, we introduced a novel metric called functional connectivity overlap ratio (FCOR), which quantifies the spatial extent of a region's connection to a given network. Using resting state functional magnetic resonance imaging data, FCOR maps were generated for several well-known large-scale resting state networks (RSNs) and used to examine the relevant associations among different RSNs, identify connector hub regions in the cerebral cortex, and elucidate the hierarchical functional organization of the brain. Constructed FCOR maps revealed a strong association among the core neurocognitive networks (default mode, salience, and executive control) as well as among primary processing networks (sensorimotor, auditory, and visual). Prominent connector hubs were identified in the bilateral middle frontal gyrus, posterior cingulate, lateral parietal, middle temporal, dorsal anterior cingulate, and anterior insula, among others, regions mostly associated with the core neurocognitive networks. Finally, clustering the whole brain using FCOR features yielded a topological organization that arranges brain regions into a hierarchy of information processing systems with the primary processing systems at one end and the heteromodal systems comprising connector hubs at the other end.