Title: Composite hybrid basis approach incorporating residual connectivity in task fMRI data
Authors: Michelle Miranda - University of Victoria (Canada) [presenting]
Jeff Morris - MD Anderson Cancer Center (United States)
Abstract: A composite hybrid basis approach is proposed to model brain spatial and temporal correlation in task fMRI data. Our model has a better detection power than the state-of-the-art methods. The greater power is obtained by borrowing both spatial and temporal information through the carefully designed basis. First, the composite hybrid basis provides a sparse spatial representation of the brain by accounting for local (within ROIs) and distant (between ROIs) correlations while yielding full Bayesian inference at the voxel and ROI level with incredible computational speed. Second, the model allows for free full Bayesian inference on the residual connectivity, which can help scientist gain insights into the underlying brain function. We apply our model to the Example Subject of the Working Memory Task of the Human Connectome Project.