Session title: New Development on Neuroimage Data Analysis
Organizer: Zhu, Hongtu (MD Anderson)
Chair: Xuan Bi(Yale)
Time: June 5th, 3:15pm – 4:45pm
Location: VEC 1303
Speech 1: A Low-Rank Multivariate General Linear Model forMulti-Subject fMRI Data and a Non-Convex Optimization Algorithm for Brain Response Comparison
Speaker: Tingting Zhang (UVA)
The focus of this paper is on evaluating brain responses to different stimuli and identifying brain regions with different responses using multi-subject, stimulus-evoked functional magnetic resonance imaging (fMRI) data. To jointly model many brain voxels’ responses to designed stimuli, we present a new low-rank multivariate general linear model (LRMGLM) for stimulus-evoked fMRI data. The new model not only is flexible to characterize variation in hemodynamic response functions (HRFs) across different regions and stimulus types, but also enables information “borrowing” across voxels and uses much fewer parameters than typical nonparametric models for HRFs. To estimate the proposed LRMGLM, we introduce a new penalized optimization function, which leads to temporally and spatially smooth HRF estimates. We develop an efficient optimization algorithm to minimize the optimization function and identify the voxels with different responses to stimuli. We show that the proposed method can outperform several existing voxel-wise methods by achieving both high sensitivity and specificity. We apply the proposed method to the fMRI data collected in an emotion study, and identify anterior dACC to have different responses to a designed threat and control stimuli.
Speech 2: Nonparametric Bayes Models of Fiber Curves Connecting Brain Regions
Speaker: Zhengwu Zhang (Rochester)
In studying structural inter-connections in the human brain, it is common to first estimate fiber bundles connecting different regions relying on diffusion MRI. These fiber bundles act as highways for neural activity. Current statistical methods reduce the rich information into an adjacency matrix, with the elements containing a count of fibers or a mean diffusion feature along the fibers. The goal of this article is to avoid discarding the rich geometric information of fibers, developing flexible models for characterizing the population distribution of fibers between brain regions of interest within and across different individuals. We start by decomposing each fiber into a rotation matrix, shape and translation from a global reference curve. These components are viewed as data lying on a product space composed of different Euclidean spaces and manifolds. To non-parametrically model the distribution within and across individuals, we rely on a hierarchical mixture of product kernels specific to the component spaces. Taking a Bayesian approach to inference, we develop efficient methods for posterior sampling. The approach automatically produces clusters of fibers within and across individuals. Applying the method to Human Connectome Project data, we find an interesting relationship between brain fiber geometry and reading ability.
Speech 3: Supervised Principal Component Regression for Functional Data with High Dimensional Predictors
Speaker: Dehan Kong (U Toronto)
Abstract: Motived by functional magnetic resonance imaging studies, we study the effect of clinical/demographic variables on the dynamic functional structures, which plays a key role in understanding brain functionality. To this end, we propose the supervised principal component regression for functional data with possibly high dimensional clinical variables. Compared with its classical counterpart, the principal component regression, the proposed methodology relies on a newly proposed integrated residual sum of squares for functional data and makes use of the association information directly. It can be formulated as a sequence of nonconvex generalized Rayleigh quotient optimization problems, which turn out to be NP-hard and thus computational intractable. Utilizing the invariance property of linear subspaces under rotations, we then reformulate the problem into a simultaneous sparse linear regression problem. Formally, we show that the reformulated problem can recover the same subspace as if the original sequence of nonconvex problems were solved. Statistically, the rate of convergence for the obtained estimators is established. Numerical studies and an application to the Human Connectome Project lend further support to the proposed methodology. (Joint work with Xinyi Zhang and Qiang Sun)