|
Activity Number:
|
26
- Imaging Speed Session
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistics in Imaging
|
|
Abstract #318577
|
|
|
Title:
|
Bayesian Inferences on Spatially Varying Correlations via Gaussian Processes
|
|
Author(s):
|
Moyan Li* and Lexin Li and Jian Kang
|
|
Companies:
|
University of Michigan, Ann Arbor and University of California, Berkeley and University of Michigan
|
|
Keywords:
|
varying coefficient model;
multimodal neuroimaging;
spatial latent factor model;
Gaussian Process;
Bayesian method
|
|
Abstract:
|
In the multimodal neuroimaging analysis, one problem of interest is to identify brain regions where different types of neural activities are strongly correlated between multiple imaging modalities. We refer to this analysis as the spatially varying correlation analysis. In this work, we propose a spatial latent factor model with the thresholded Gaussian Process (TGP) prior for spatially varying correlation analysis. Compared to the existing methods, the proposed Bayesian method is more powerful and has the ability to incorporate prior knowledge of brain and to appropriately quantify the region selection uncertainty. We also develop an efficient sampling algorithm for posterior computation. Compared to the gradient based MCMC methods, our algorithm is more computationally efficient with a fast convergence. We illustrate our method via extensive simulation studies and analysis of multimodal imaging data in a large scale neuroimaging study.
|
Authors who are presenting talks have a * after their name.
