Related Experiment Video
Updated: Sep 17, 2025
06:26
Meta-analysis of Voxel-Based Neuroimaging Studies using Seed-based d Mapping with Permutation of Subject Images SDM-PSI
Published on: November 27, 2019
73.2K
Gray Matter Volume Abnormalities in Schizophrenia: Comparisons Between P-Value and Effect Size Inference Frameworks
Xin Li1, Wenshuang Zhu1, Zhen Zhao1
1Department of Radiology, Tianjin Key Lab of Functional Imaging, Tianjin Institute of Radiology and State Key Laboratory of Experimental Hematology, Tianjin Medical University General Hospital, Tianjin 300052, China.
Schizophrenia Bulletin
|June 27, 2025
Summary
Effect size (ES) inference reliably identifies brain abnormalities in schizophrenia, outperforming P-value methods. Image-based meta-analysis (IBMA) shows superior detection of structural changes in multi-center studies.
Area of Science:
- Neuroimaging
- Psychiatric Disorders
- Statistical Analysis
Background:
- Identifying generalizable brain imaging markers for schizophrenia is challenging due to data aggregation methods and p-hacking.
- Large multi-center datasets require robust statistical approaches to overcome variability.
Purpose of the Study:
- To compare the efficacy of effect size (ES) inference versus P-value inference for identifying brain abnormalities in schizophrenia.
- To evaluate different statistical aggregation methods: Mega-analysis (Mega), Image-Based Meta-analysis (IBMA), and Coordinate-Based Meta-analysis (CBMA).
Main Methods:
- Voxel-wise gray matter volume (GMV) differences were analyzed using individual data from 976 schizophrenia patients and 801 controls across 16 datasets.
- Published coordinates data from 103 studies (5151 patients, 5438 controls) were analyzed using Mega-analysis, IBMA, and CBMA under P-value and ES inference frameworks.
Main Results:
- P-value Mega-analysis identified widespread GMV abnormalities (94.85% of voxels) highly sensitive to sample size.
- Effect size Mega-analysis detected core abnormalities in fewer voxels (24.63%) but showed greater resistance to sample size.
- ES-based IBMA and CBMA outperformed P-value methods, with IBMA demonstrating comparable performance to Mega-analysis and superior results to CBMAs.
Conclusions:
- Effect size inference offers advantages for statistical aggregation in multi-center neuroimaging studies.
- Image-Based Meta-analysis (IBMA) shows significant potential for reliably detecting brain structural abnormalities in schizophrenia.
