Neuroimaging Meta Regression for Coordinate Based Meta Analysis Data with a Spatial Model

Abstract

Coordinate-based meta-analysis combines evidence from a collection of Neuroimaging studies to estimate brain activation. In such analyses, a key practical challenge is to find a computationally efficient approach with good statistical interpretability to model the locations of activation foci. In this article, we propose a generative coordinate-based meta-regression (CBMR) framework to approximate smooth activation intensity function and investigate the effect of study-level covariates (e.g., year of publication, sample size). We employ spline parameterization to model spatial structure of brain activation and consider four stochastic models for modelling the random variation in foci. To examine the validity of CBMR, we estimate brain activation on $20$ meta-analytic datasets, conduct spatial homogeneity tests at voxel level, and compare to results generated by existing kernel-based approaches.