Bayesian spatial transformation models with applications in neuroimaging data

TL;DR

This paper introduces spatial transformation models (STMs) for neuroimaging data that effectively handle non-Gaussian distributions and spatial dependencies, improving the detection of brain region changes.

Contribution

The paper develops a novel class of spatial transformation models combining Box-Cox transformations and Gaussian Markov Random Fields for neuroimaging analysis.

Findings

STM outperforms voxel-wise linear models in pattern recovery

Successfully identified brain regions with morphological changes in ADHD children

Efficient MCMC algorithm enables practical application

Abstract

The aim of this paper is to develop a class of spatial transformation models (STM) to spatially model the varying association between imaging measures in a three-dimensional (3D) volume (or 2D surface) and a set of covariates. Our STMs include a varying Box-Cox transformation model for dealing with the issue of non-Gaussian distributed imaging data and a Gaussian Markov Random Field model for incorporating spatial smoothness of the imaging data. Posterior computation proceeds via an efficient Markov chain Monte Carlo algorithm. Simulations and real data analysis demonstrate that the STM significantly outperforms the voxel-wise linear model with Gaussian noise in recovering meaningful geometric patterns. Our STM is able to reveal important brain regions with morphological changes in children with attention deficit hyperactivity disorder.