CBMR: Coordinate-based meta-regression for group and covariate inference

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Statistical Analysis

Background:

• Coordinate-based meta-analysis (CBMA) identifies brain activation patterns across studies.
• Comparing activation foci distributions between study groups in CBMA is challenging, often necessitating balanced sample sizes.

Purpose of the Study:

• To introduce a flexible multi-group coordinate-based meta-regression (CBMR) framework.
• To enable robust comparisons of brain activation patterns across multiple neuroimaging study groups, irrespective of sample size balance.

Main Methods:

• Developed a generative spline-based spatial model for CBMR.
• Incorporated a roughness penalty for flexible control over model smoothness.
• Evaluated the framework using simulations and real neuroimaging data.

Main Results:

• Parametric inference is valid for groups with at least 200 foci.
• Sparser datasets require inference via parametric bootstrap for accurate results.
• The CBMR framework demonstrates flexibility and validity in multi-group analyses.

Conclusions:

• The novel CBMR framework overcomes limitations of traditional CBMA for multi-group comparisons.
• The method is freely available as a NiMARE module, facilitating its use in functional MRI meta-analyses.
• Enables flexible meta-regression and inference for diverse coordinate-based meta-analytic datasets.