Thursday, May 26, 2022 • 4:50–6:05 pm (CT) • Light Hall, Room 512
Organizer: Rajarshi Guhaniyogi, Texas A&M University
Chair: Ben Risk, Emory University
A variational Bayesian approach to identifying whole-brain directed networks with fMRI data
Tingting Zhang, University of Pittsburgh
Yaotian Wang, University of Pittsburgh
The brain is a high-dimensional directed network system as it consists of many regions as network nodes that exert influence on each other. The directed influence exerted by one region on another is referred to as directed connectivity. We aim to reveal whole-brain directed networks based on resting-state functional magnetic resonance imaging (fMRI) data of many subjects. However, it is both statistically and computationally challenging to produce scientifically meaningful estimates of whole-brain directed networks. To address the statistical modeling challenge, we assume modular brain networks, which reflect functional specialization and functional integration of the brain. We address the computational challenge by developing a variational Bayesian method to estimate the new model. We apply our method to resting-state fMRI data of many subjects and identify modules and directed connections in whole-brain directed networks. The identified modules are accordant with functional brain systems specialized for different functions. We also detect directed connections between functionally specialized modules, which is not attainable by existing network methods based on functional connectivity. In summary, this paper presents a new computationally efficient and flexible method for directed network studies of the brain as well as new scientific findings regarding the functional organization of the human brain.
Tangent functional canonical correlation analysis for densities and shapes, with applications to multimodal imaging data
Sebastian Kurtek, The Ohio State University
Min Ho Cho, Inha University
Karthik Bharath, University of Nottingham
It is quite common for functional data arising from imaging data to assume values in infinite-dimensional manifolds. Uncovering associations between two or more such nonlinear functional data extracted from the same object across medical imaging modalities can assist development of personalized treatment strategies. We propose a method for canonical correlation analysis between paired probability densities or shapes of closed planar curves, routinely used in biomedical studies, which combines a convenient linearization and dimension reduction of the data using tangent space coordinates. Leveraging the fact that the corresponding manifolds are submanifolds of unit Hilbert spheres, we describe how finite-dimensional representations of the functional data objects can be easily computed, which then facilitates use of standard multivariate canonical correlation analysis methods. We further construct and visualize canonical variate directions directly on the space of densities or shapes. Utility of the method is demonstrated through numerical simulations and performance on a magnetic resonance imaging dataset of glioblastoma multiforme brain tumors.
A Bayesian covariance–based clustering for high-dimensional tensors
Aaron Wolfe Scheffler, University of California, San Francisco
Rene Gutierrez Marquez, University of California, Santa Cruz
Rajarshi Guhaniyogi, Texas A&M University
Clustering of high-dimensional tensors with limited sample size has become prevalent in a variety of application areas. Existing Bayesian model-based clustering of tensors yields less accurate clusters when the tensor dimensions are sufficiently large, sample size is low and clusters of tensors mainly reveal difference in their variability. We propose a novel clustering technique for high-dimensional tensors with limited sample size when the clusters show difference in their covariances, rather than in their means. The proposed approach constructs several matrices from a tensor to adequately estimate its variability along different modes and implements a model-based approximate Bayesian clustering algorithm with the matrices thus constructed, in place with the original tensor data. Although some information in the data is discarded, we gain substantial computational efficiency and accuracy in clustering. A simulation study assesses the proposed approach along with its competitors. The proposed methodology is applied to resting-state EEG data collected on children with autism spectrum disorder (ASD) and provides novel insights into neurodevelopmental heterogeneity within ASD.