Easy to interpret Coordinate Based Meta-Analysis of neuroimaging studies: Analysis of Brain Coordinates (ABC)

Abstract

AbstractFunctional MRI and voxel-based morphometry (VBM) are important approaches to testing hypotheses in neuroscience, helping us to understand neurological disease, and brain function and development. However, they are technically challenging with no one optimal generalisable method, and the multiple popular techniques have been shown to produce different results. Furthermore, results may be sensitive to settings, such as smoothing or statistical thresholding, that can be difficult to optimise per hypothesis. It is useful, therefore, to be able to meta-analyse published results from such studies that tested a similar hypothesis potentially using different analysis methods, scanners, and protocols as well as different subjects. Coordinate based meta-analysis (CBMA) offers this using only commonly reported summary results. It is the aim of CBMA to find those results that indicate replicable effects across studies. However, just like the multiple analysis methods offered for neuroimaging, there are now multiple CBMA algorithms each with specific features and empirical parameters/assumptions. Results derived from CBMA are inevitably conditional on the algorithm used, so conclusions are clearer when the analysis approach is easy to understand. With this in mind a new CBMA method (Analysis of Brain Coordinates; ABC) is presented, with the aim of being easy to interpret by eliminating empirical assumptions where possible and by relating statistical thresholding directly to replication of effect.