BLMM: Parallelised computing for big linear mixed models
Within neuroimaging large-scale, shared datasets are becoming increasingly commonplace, challenging existing tools both in terms of overall scale and complexity of the study designs. As sample sizes grow, researchers are presented with new opportunities to detect and account for grouping factors and...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2022-12-01
|
| Series: | NeuroImage |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S1053811922008503 |
| _version_ | 1811178660813602816 |
|---|---|
| author | Thomas Maullin-Sapey Thomas E. Nichols |
| author_facet | Thomas Maullin-Sapey Thomas E. Nichols |
| author_sort | Thomas Maullin-Sapey |
| collection | DOAJ |
| description | Within neuroimaging large-scale, shared datasets are becoming increasingly commonplace, challenging existing tools both in terms of overall scale and complexity of the study designs. As sample sizes grow, researchers are presented with new opportunities to detect and account for grouping factors and covariance structure present in large experimental designs. In particular, standard linear model methods cannot account for the covariance and grouping structures present in large datasets, and the existing linear mixed models (LMM) tools are neither scalable nor exploit the computational speed-ups afforded by vectorisation of computations over voxels. Further, nearly all existing tools for imaging (fixed or mixed effect) do not account for variability in the patterns of missing data near cortical boundaries and the edge of the brain, and instead omit any voxels with any missing data. Yet in the large-n setting, such a voxel-wise deletion missing data strategy leads to severe shrinkage of the final analysis mask. To counter these issues, we describe the “Big” Linear Mixed Models (BLMM) toolbox, an efficient Python package for large-scale fMRI LMM analyses. BLMM is designed for use on high performance computing clusters and utilizes a Fisher Scoring procedure made possible by derivations for the LMM Fisher information matrix and score vectors derived in our previous work, Maullin-Sapey and Nichols (2021). |
| first_indexed | 2024-04-11T06:22:09Z |
| format | Article |
| id | doaj.art-c09cbaf89f8744dcb51b425976bc8d66 |
| institution | Directory Open Access Journal |
| issn | 1095-9572 |
| language | English |
| last_indexed | 2024-04-11T06:22:09Z |
| publishDate | 2022-12-01 |
| publisher | Elsevier |
| record_format | Article |
| series | NeuroImage |
| spelling | doaj.art-c09cbaf89f8744dcb51b425976bc8d662022-12-22T04:40:32ZengElsevierNeuroImage1095-95722022-12-01264119729BLMM: Parallelised computing for big linear mixed modelsThomas Maullin-Sapey0Thomas E. Nichols1Corresponding author.; Big Data Institute, Li Ka Shing Centre for Health Information and Discovery, Nuffield Department of Population Health, University of Oxford, Oxford, UKBig Data Institute, Li Ka Shing Centre for Health Information and Discovery, Nuffield Department of Population Health, University of Oxford, Oxford, UK; Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UKWithin neuroimaging large-scale, shared datasets are becoming increasingly commonplace, challenging existing tools both in terms of overall scale and complexity of the study designs. As sample sizes grow, researchers are presented with new opportunities to detect and account for grouping factors and covariance structure present in large experimental designs. In particular, standard linear model methods cannot account for the covariance and grouping structures present in large datasets, and the existing linear mixed models (LMM) tools are neither scalable nor exploit the computational speed-ups afforded by vectorisation of computations over voxels. Further, nearly all existing tools for imaging (fixed or mixed effect) do not account for variability in the patterns of missing data near cortical boundaries and the edge of the brain, and instead omit any voxels with any missing data. Yet in the large-n setting, such a voxel-wise deletion missing data strategy leads to severe shrinkage of the final analysis mask. To counter these issues, we describe the “Big” Linear Mixed Models (BLMM) toolbox, an efficient Python package for large-scale fMRI LMM analyses. BLMM is designed for use on high performance computing clusters and utilizes a Fisher Scoring procedure made possible by derivations for the LMM Fisher information matrix and score vectors derived in our previous work, Maullin-Sapey and Nichols (2021).http://www.sciencedirect.com/science/article/pii/S1053811922008503 |
| spellingShingle | Thomas Maullin-Sapey Thomas E. Nichols BLMM: Parallelised computing for big linear mixed models NeuroImage |
| title | BLMM: Parallelised computing for big linear mixed models |
| title_full | BLMM: Parallelised computing for big linear mixed models |
| title_fullStr | BLMM: Parallelised computing for big linear mixed models |
| title_full_unstemmed | BLMM: Parallelised computing for big linear mixed models |
| title_short | BLMM: Parallelised computing for big linear mixed models |
| title_sort | blmm parallelised computing for big linear mixed models |
| url | http://www.sciencedirect.com/science/article/pii/S1053811922008503 |
| work_keys_str_mv | AT thomasmaullinsapey blmmparallelisedcomputingforbiglinearmixedmodels AT thomasenichols blmmparallelisedcomputingforbiglinearmixedmodels |