Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
[Preprint]. 2025 Jun 17:2025.06.16.657725.
doi: 10.1101/2025.06.16.657725.

Image-Based Meta- and Mega-Analysis (IBMMA): A Unified Framework for Large-Scale, Multi-Site, Neuroimaging Data Analysis

Nick Steele  1   2 Rajendra A Morey  1   2 Ahmed Hussain  1   2 Courtney Russell  1   2 Benjamin Suarez-Jimenez  3 Elena Pozzi  4   5 Hadis Jameei  4 Lianne Schmaal  4   5 Ilya M Veer  6 Lea Waller  7 Neda Jahanshad  8 Sophia I Thomopoulos  8 Lauren E Salminen  8 Miranda Olff  9   10 Jessie L Frijling  11   9 Dick J Veltman  9   12 Saskia B J Koch  9   13 Laura Nawijn  9 Mirjam van Zuiden  9   14 Li Wang  15   16 Ye Zhu  15   16 Gen Li  15   16 Dan J Stein  17 Jonathan Ipser  17 Yuval Neria  18   19 Xi Zhu  18   19 Orren Ravid  19 Sigal Zilcha-Mano  20 Amit Lazarov  21 Ashley A Huggins  22 Jennifer S Stevens  23 Kerry Ressler  24   25 Tanja Jovanovic  26   23 Sanne J H van Rooij  23 Negar Fani  23 Sven C Mueller  27 Anna R Hudson  27 Judith K Daniels  28 Anika Sierk  29 Antje Manthey  29 Henrik Walter  29 Nic J A van der Wee  30   31 Steven J A van der Werff  30   31 Robert R J M Vermeiren  32 Christian Schmahl  33   34 Julia I Herzog  33   34 Ivan Rektor  35 Pavel Říha  36   35 Milissa L Kaufman  25   37 Lauren A M Lebois  25   24 Justin T Baker  25   38 Isabelle M Rosso  25   39 Elizabeth A Olson  25   39 Anthony King  40 Israel Liberzon  41 Michael Angstadt  40 Nicholas D Davenport  42   43 Seth G Disner  42   43 Scott R Sponheim  42 Thomas Straube  44 David Hofmann  44 Guangming Lu  45 Rongfeng Qi  46 Xin Wang  47 Austin Kunch  47 Hong Xie  47 Yann Quidé  48   49 Wissam El-Hage  50 Shmuel Lissek  51 Hannah Berg  51 Steven E Bruce  52 Josh Cisler  53 Marisa Ross  54 Ryan J Herringa  55 Daniel W Grupe  56 Jack B Nitschke  57 Richard J Davidson  56   57   58 Christine Larson  59 Terri A deRoon-Cassini  60   61 Carissa W Tomas  62   61 Jacklynn M Fitzgerald  63 Jeremy Elman  64   65 Matthew Panizzon  64 Carol E Franz  66   67 Michael J Lyons  68 William S Kremen  66   67   69 Brandee Feola  70 Jennifer U Blackford  71 Bunmi O Olatunji  72 Geoffrey May  73   74   75 Steven M Nelson  75   74   73   76 Evan M Gordon  77 Chadi G Abdallah  78   79 Ruth Lanius  80   81 Maria Densmore  81 Jean Théberge  81 Richard W J Neufeld  81 Paul M Thompson  8 Delin Sun  82   1   2
Affiliations

Image-Based Meta- and Mega-Analysis (IBMMA): A Unified Framework for Large-Scale, Multi-Site, Neuroimaging Data Analysis

Nick Steele et al. bioRxiv. .

Update in

  • Image-Based Meta- and Mega-Analysis (IBMMA): A Unified Framework for Large-Scale, Multi-Site, Neuroimaging Data Analysis.
    Steele N, Huggins AA, Morey RA, Hussain A, Russell C, Suarez-Jimenez B, Pozzi E, Jameei H, Schmaal L, Veer IM, Waller L, Jahanshad N, Thomopoulos SI, Salminen LE, Olff M, Frijling JL, Veltman DJ, Koch SBJ, Nawijn L, van Zuiden M, Wang L, Zhu Y, Li G, Stein DJ, Ipser J, Neria Y, Zhu X, Ravid O, Zilcha-Mano S, Lazarov A, Stevens JS, Ressler K, Jovanovic T, van Rooij SJH, Fani N, Mueller SC, Hudson AR, Daniels JK, Sierk A, Manthey A, Walter H, van der Wee NJA, van der Werff SJA, Vermeiren RRJM, Schmahl C, Herzog JI, Rektor I, Říha P, Kaufman ML, Lebois LAM, Baker JT, Rosso IM, Olson EA, King A, Liberzon I, Angstadt M, Davenport ND, Disner SG, Sponheim SR, Straube T, Hofmann D, Lu G, Qi R, Wang X, Kunch A, Xie H, Quidé Y, El-Hage W, Lissek S, Berg H, Bruce SE, Cisler J, Ross M, Herringa RJ, Grupe DW, Nitschke JB, Davidson RJ, Larson C, deRoon-Cassini TA, Tomas CW, Fitzgerald JM, Elman J, Panizzon M, Franz CE, Lyons MJ, Kremen WS, Feola B, Blackford JU, Olatunji BO, May G, Nelson SM, Gordon EM, Abdallah CG, Lanius R, Densmore M, Théberge J, Neufeld RWJ, Thompson PM, Sun D. Steele N, et al. Neuroimage. 2025 Oct 23:121554. doi: 10.1016/j.neuroimage.2025.121554. Online ahead of print. Neuroimage. 2025. PMID: 41138791

Abstract

The increasing scale and complexity of neuroimaging datasets aggregated from multiple study sites present substantial analytic challenges, as existing statistical analysis tools struggle to handle missing voxel-data, suffer from limited computational speed and inefficient memory allocation, and are restricted in the types of statistical designs they are able to model. We introduce Image-Based Meta- & Mega-Analysis (IBMMA), a novel software package implemented in R and Python that provides a unified framework for analyzing diverse neuroimaging features, efficiently handles large-scale datasets through parallel processing, offers flexible statistical modeling options, and properly manages missing voxel-data commonly encountered in multi-site studies. IBMMA produced stronger effect sizes and revealed findings in brain regions that traditional software overlooked due to missing voxel-data resulting in gaps in brain coverage. IBMMA has the potential to accelerate discoveries in neuroscience and enhance the clinical utility of neuroimaging findings.

Keywords: Big Data; Mega-analysis; Meta-analysis; Neuroimaging; PTSD; Resting-state fMRI.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.. Schematic of IBMMA Workflow.
Neuroimaging data from multiple subjects are flattened and segmented into memory-compatible subject-by-feature matrices. Statistical models are applied in parallel across features, and the resulting outputs are extracted and reconstructed to match the original data dimensions.
Figure 2.
Figure 2.. Steps to Run IBMMA.
A) IBMMA expects neuroimaging data to be organized in site-specific subdirectories. Within each subdirectory, neuroimaging files should be named starting with the subject ID followed by a naming pattern that is consistent across all files. Within the clinical data file, IBMMA expects a fID column whose elements are composed of site ID and subject ID. B) IBMMA requires the user to first input clinical and neuroimaging data file paths and model specifications in the para_path.xlsx file. C) IBMMA can be run from the command line, either locally or via a HPC, by running the ibmma.py script. D) Statistical results will be automatically generated once the script is done running and can be viewed using any neuroimaging viewer (e.g., FSLeyes, freeview, MRIcron). An HTML webpage is also generated to view results.
Figure 3.
Figure 3.. IBMMA Outputs.
A) Total run time, CPU time, and maximum virtual memory usage of each processing step of IBMMA’s mega-analysis using a sample of n = 2,611. B) IBMMA generated whole-brain maps of various statistical and model fit measures. C) An automatically generated HTML webpage provided result tables and visualizations of neuroimaging findings. All tables were also saved as CSV files, and all significant voxel clusters were saved as png images.
Figure 4.
Figure 4.
A) Brain maps displaying the proportion of subjects with missing data at each voxel of the brain in the ENIGMA-PTSD dataset (n = 3,193). B) The sensitivity of IBMMA at a 5% false positive rate (FPR) calculated from a simulated dataset given the observed missing rates at each voxel of the brain. C) Receiver operator characteristic (ROC) curves at different levels of subject missingness in increments of 5%. D) Performance of IBMMA, measured by area under the curve (AUC) and sensitivity at a 5% FPR, across varying levels of subject missingness.
Figure 5.
Figure 5.
Seed-based functional connectivity results for the left thalamus using IBMMA, FSL, and conjunction analyses. A) IBMMA accounts for missing voxel-data by only using subjects with non-missing data for a given voxel, while FSL only analyzes voxels with complete data. Left: Number of participants with non-missing voxel-data used by IBMMA for each voxel. Right: The colored regions represent voxels that were included in the FSL analysis, while non-colored regions were excluded due to one or more participants having missing voxel-data. B, C, D) Significant associations found by IBMMA (left) and FSL (right). Z-statistic maps are thresholded at Z ≥ 3.1 (uncorrected). In the conjunction analysis panels (bottom), red voxels indicate regions identified as significant by both IBMMA and FSL, while purple (IBMMA only) and blue (FSL only) voxels indicate areas of disagreement between the two software. B) Significant negative associations with age. C) Significant associations with sex (female > male). D) Significant association with PTSD diagnosis (cases > controls).

References

    1. Cox R. W. (1996). AFNI: Software for Analysis and Visualization of Functional Magnetic Resonance Neuroimages. Computers and Biomedical Research, 29(3), 162–173. 10.1006/cbmr.1996.0014 - DOI - PubMed
    1. Esteban O., Markiewicz C. J., Blair R. W., Moodie C. A., Isik A. I., Erramuzpe A., Kent J. D., Goncalves M., DuPre E., Snyder M., Oya H., Ghosh S. S., Wright J., Durnez J., Poldrack R. A., & Gorgolewski K. J. (2019). fMRIPrep: A robust preprocessing pipeline for functional MRI. Nature Methods, 16(1), 111–116. 10.1038/s41592-018-0235-4 - DOI - PMC - PubMed
    1. Hall J. A., & Rosenthal R. (2018). Choosing between random effects models in meta-analysis: Units of analysis and the generalizability of obtained results. Social and Personality Psychology Compass, 12(10), e12414. 10.1111/spc3.12414 - DOI
    1. Jenkinson M., Beckmann C. F., Behrens T. E. J., Woolrich M. W., & Smith S. M. (2012). FSL. NeuroImage, 62(2), 782–790. 10.1016/j.neuroimage.2011.09.015 - DOI - PubMed
    1. Maullin-Sapey T., & Nichols T. E. (2022). BLMM: Parallelised computing for big linear mixed models. Neuroimage, 264, 119729. 10.1016/j.neuroimage.2022.119729 - DOI - PMC - PubMed

Publication types

LinkOut - more resources