Related Experiment Videos
Coordinate based random effect size meta-analysis of neuroimaging studies.
C R Tench1, Radu Tanasescu2, C S Constantinescu1
1Division of Clinical Neurosciences, Clinical Neurology, University of Nottingham, Queen's Medical Centre, Nottingham, UK.
Neuroimage
|April 9, 2017
Summary
Coordinate-based meta-analysis (CBMA) improves neuroimaging study power. The new ClusterZ algorithm performs random effects meta-analysis and meta-regression on coordinates and effect sizes, enhancing statistical rigor and controlling errors.
Area of Science:
- Neuroimaging
- Neuroscience
- Biostatistics
Background:
- Low statistical power in neuroimaging studies hinders interpretation.
- Coordinate-based meta-analysis (CBMA) aggregates findings from published studies.
- Existing CBMA methods rely on coordinate density for significance.
Purpose of the Study:
- Introduce a novel coordinate-based random effects meta-analysis and meta-regression method.
- Develop an algorithm (ClusterZ) to analyze coordinates and effect sizes.
- Improve statistical rigor in neuroimaging meta-analyses.
Main Methods:
- The ClusterZ algorithm analyzes coordinates and reported t/Z scores, standardized by subject number.
- Statistical significance is determined by random effects meta-analysis of reported effects, accounting for data censoring.
- Type 1 error is controlled using the false cluster discovery rate (FCDR).
Main Results:
- ClusterZ was validated using simulated data and real neuroimaging datasets.
- Demonstrated effectiveness on grey matter loss in multiple sclerosis and pain perception studies.
- The method provides robust statistical significance determination beyond coordinate density.
Conclusions:
- ClusterZ offers a statistically rigorous approach to coordinate-based meta-analysis.
- The method effectively mitigates issues associated with low power in neuroimaging.
- Software implementation is freely available for broader research application.