Easy to interpret coordinate based meta-analysis of neuroimaging studies: Analysis of brain coordinates (ABC)

• Published in: Journal of neuroscience methods 2022-04, Vol.372, p.109556-109556, Article 109556
• Main Authors: Tench, CR, Tanasescu, R., Constantinescu, CS, Auer, DP, Cottam, WJ
• Format: Article
• Language: English
• Subjects: Algorithms; Brain - diagnostic imaging; Functional MRI; Gray Matter - diagnostic imaging; Magnetic Resonance Imaging; Meta-analysis; Neuroimaging; Voxel-based morphometry
• ISSN: 0165-0270; 1872-678X
• DOI: 10.1016/j.jneumeth.2022.109556

Functional MRI and voxel-based morphometry are important in neuroscience. They are technically challenging with no globally optimal analysis method, and the multiple approaches have been shown to produce different results. It is useful to be able to meta-analyse results from such studies that tested a similar hypothesis potentially using different analysis methods. The aim is to identify replicable results and infer hypothesis specific effects. Coordinate based meta-analysis (CBMA) offers this, but the multiple algorithms can produce different results, making interpretation conditional on the algorithm. Here a new model based CBMA algorithm, Analysis of Brain Coordinates (ABC), is presented. ABC aims to be simple to understand by avoiding empirical elements where possible and by using a simple to interpret statistical threshold, which relates to the primary aim of detecting replicable effects. ABC is compared to both the most used and the most recently developed CBMA algorithms, by reproducing a published meta-analysis of localised grey matter changes in schizophrenia. There are some differences in results and the type of data that can be analysed, which are related to the algorithm specifics. Compared to other algorithms ABC eliminates empirical elements where possible and uses a simple to interpret statistical threshold. There may be no optimal way to meta-analyse neuroimaging studies using CBMA. However, by eliminating some empirical elements and relating the statistical threshold directly to the aim of finding replicable effects, ABC makes the impact of the algorithm on any conclusion easier to understand. •CBMA algorithm with fewer empirical components.•Easy to interpret statistical threshold.•Computationally efficient.•Free to use software.