B0384
Title: Network-level analysis for connectome-wide association studies
Authors: Muriah Wheelock - Washington University in St. Louis (United States) [presenting]
Abstract: Methods that enable quantification of brain connectivity-behaviour relationships are crucial for understanding fundamental biological mechanisms underlying behavioural and clinical outcomes. While most human functional neuroimaging software has focused on contiguous voxel extents for controlling the false positive rate, a novel method is developed based on statistics used in genome-wide association studies that do not assume brain regions are spatially contiguous. Instead, network-level analysis (NLA) software leverages the systems-level organization of the human brain to determine significant behavioural or clinical outcome associations while probing all possible functional connections, (i.e., connectome-wide associations). Specifically, NLA first fits statistical models at the individual functional connection (i.e., edge) level to model brain associations with individual variability in behaviour or differences between groups. Then, NLA fits enrichment statistics at the systems (i.e., network) level to assess whether associations with functional connections are stronger within certain brain systems relative to others. The significance of brain-behaviour associations at the network level is determined through permutation testing in which the behaviour or group labels are permuted thousands of times to form a null distribution. Finally, a variety of publication quality figures are available within the software.
