Identifying consistent brain networks via maximizing predictability of functional connectome from structural connectome

• Published in: 2013 IEEE 10th International Symposium on Biomedical Imaging pp. 978 - 981
• Main Authors: Hanbo Chen, Kaiming Li, Dajiang Zhu, Tianming Liu
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.04.2013
• Subjects: Brain Functional/Structural Connectomes; Clustering algorithms; Diffusion tensor imaging; Multi-view Spectral Clustering; Neuroscience; Optimization; Prediction algorithms; Visualization
• ISBN: 1467364568; 9781467364560
• ISSN: 1945-7928
• DOI: 10.1109/ISBI.2013.6556640

Recent studies have suggested that structural brain connectivity is strongly correlated with functional connectivity. However, the relationship between structural and functional connectivity at the whole brain connectome scale has been rarely explored. This paper presents a novel framework to infer brain networks that are consistent across multiple neuroimaging modalities and across individuals at the connectome scale. Our basic premise is that the predictability of functional connectivity from structural connectivity within each brain network should be maximized, which is formulated by and solved via a novel feedback-regulated multi-view spectral clustering algorithm. We applied and tested the proposed algorithm on the multimodal structural and functional brain connectomes of 50 healthy subjects, and obtained promising results. Our validation experiments demonstrated that the derived brain networks are in agreement with current neuroscience knowledge and offer novel insights into the close relationship between brain structure and function at the connectome scale.