OHBM 2017: Practical intensity based meta-analysis
1 like348 views
The document outlines a meta-analysis educational course led by Camille Maumet focused on neuroimaging techniques, specifically discussing coordinate-based and image-based meta-analyses. It highlights the advantages of image-based meta-analysis when data is available and describes the practical challenges of implementing the gold standard mixed-effects GLM method. Additionally, it provides insights into statistical methods for meta-analysis and the importance of data sharing tools like Neurovault.
![Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 18
• Contrast Estimation
– Linear combination of parameter estimates
– Final statistics invariant to scale
• e.g. [ 1 1 1 1 ] gives same T’s & P’s as [ ¼ ¼ ¼ ¼ ]
Units depend on contrast vector
– Rule for contrasts to preserve units
• Positive elements sum to 1
• Negative elements sum to -1
Pre-processed
data
Data scaling
Scaled
pre-proc. data
Model
parameter
estimation Parameter
estimates
Contrast
estimation Contrast and
std. err. maps
Contrast
estimation
Units of contrast estimates](https://image.slidesharecdn.com/ohbm2017-practicalintensitybasedmeta-analysis-170625201214/85/OHBM-2017-Practical-intensity-based-meta-analysis-18-320.jpg)
![Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 20
3dMEMA_result+tlrc.BRIK[[0]]
[from contrast & stat maps]
Which images for IBMA?
Contrast & std. err.
maps
Statistic map
E.g. Z-map
Contrast map
SPM FSL AFNI
con_0001.nii
[SPM.mat]
cope1.nii
varcope1.nii (squared)
3dMEMA_result+tlrc.BRIK[[1]]spmT_0001.nii tstat1.nii.gz
zstat1.nii.gz
3dMEMA_result+tlrc.BRIK[[0]]con_0001.nii cope1.nii](https://image.slidesharecdn.com/ohbm2017-practicalintensitybasedmeta-analysis-170625201214/85/OHBM-2017-Practical-intensity-based-meta-analysis-20-320.jpg)
OHBM 2017: Practical intensity based meta-analysis
- 1. Practical intensity-based meta-analysis Camille Maumet OHBM Neuroimaging Meta-Analysis Educational course, 25 June 2017 Coordinate-based meta-analysis Image-based meta-analysis
- 3. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 3 Neuroimaging meta-analyses Acquisition Analysis Experiment Raw data Results Acquisition Analysis Experiment Raw data Results … Publication Publication Paper Paper
- 4. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 4 Acquisition Analysis Experiment Raw data Results Acquisition Analysis Experiment Raw data Results … Publication Publication Paper Paper Coordinate-based meta-analysis Coordinate-based meta-analysis Neuroimaging meta-analyses
- 5. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 5 Image-based meta-analysis Shared results Data sharing Acquisition Analysis Experiment Raw data Results Acquisition Analysis Experiment Raw data Results … Publication Publication Paper Paper Coordinate-based meta-analysis Coordinate-based meta-analysis Image-based meta-analysis Neuroimaging meta-analyses
- 6. How to perform an image-based meta-analysis?
- 7. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 7 Inference Detections (subject-level) Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Image-based meta-analysis
- 8. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 8 Inference Detections (subject-level) Inference Detections (subject-level) Image-based meta-analysis Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps …
- 9. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 9 Inference Detections (study-level) Image-based meta-analysis Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps
- 10. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 10 Inference Detections (study-level) Inference Detections (study-level) Image-based meta-analysis Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps
- 11. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 11 Image-based meta-analysis Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps Model fitting and estimation Contrast and std. err. maps Inference Detections (meta-analysis)
- 12. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 12 Inference Detections (subject-level) Inference Detections (subject-level) Inference Detections (study-level) Inference Detections (study-level) Meta-analysis levelStudy levelSubject level Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Model fitting and estimation Contrast and std. err. maps Model fitting and estimation Contrast and std. err. maps Inference Detections (meta-analysis) Image-based meta-analysis
- 13. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 13 • Gold standard: Third-level Mixed-Effects GLM • Requirements – study-level Contrast estimates and Standard error maps. – Same units Contrast and std. err. maps Image-based meta-analysis
- 14. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 14 Units of contrast estimates Pre-processed data Model fitting and estimation Contrast and std. err. maps
- 15. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 15 Pre-processed data Model fitting and estimation Contrast and std. err. maps Pre-processed data Data scaling Scaled pre-proc. data Model parameter estimation Parameter estimates Contrast estimation Contrast and std. err. maps Units of contrast estimates
- 16. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 16 Units depend on mean estimation and scaling target. Pre-processed data Data scaling Scaled pre-proc. data Model parameter estimation Parameter estimates Contrast estimation Contrast and std. err. maps Data scaling Units of contrast estimates
- 17. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 17 Y = β + Units depend on scaling of explanatory variables Pre-processed data Data scaling Scaled pre-proc. data Model parameter estimation Parameter estimates Contrast estimation Contrast and std. err. maps Model parameter estimation Units of contrast estimates
- 18. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 18 • Contrast Estimation – Linear combination of parameter estimates – Final statistics invariant to scale • e.g. [ 1 1 1 1 ] gives same T’s & P’s as [ ¼ ¼ ¼ ¼ ] Units depend on contrast vector – Rule for contrasts to preserve units • Positive elements sum to 1 • Negative elements sum to -1 Pre-processed data Data scaling Scaled pre-proc. data Model parameter estimation Parameter estimates Contrast estimation Contrast and std. err. maps Contrast estimation Units of contrast estimates
- 19. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 19 • Gold standard: • But… – Units will depend on: • The scaling of the data (subject-level) • The scaling of the predictor(s) (subject- and study-level) • The scaling of the contrast (subject- and study-level). – Contrast estimates and standard error maps are rarely shared… Third-level Mixed-Effects GLM Units of contrast estimates
- 20. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 20 3dMEMA_result+tlrc.BRIK[[0]] [from contrast & stat maps] Which images for IBMA? Contrast & std. err. maps Statistic map E.g. Z-map Contrast map SPM FSL AFNI con_0001.nii [SPM.mat] cope1.nii varcope1.nii (squared) 3dMEMA_result+tlrc.BRIK[[1]]spmT_0001.nii tstat1.nii.gz zstat1.nii.gz 3dMEMA_result+tlrc.BRIK[[0]]con_0001.nii cope1.nii
- 21. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 21 • Fisher's – Sum of −log P-values (from T/Z’s converted to P’s) • Stouffer’s – Average Z, rescaled to N(0,1) • “Stouffer's Random Effects (RFX)” – Submit Z’s to one-sample t-test IBMA on Z maps (Slide adapted from Thomas Nichols, OHBM 2015) Statistic map E.g. Z-map
- 22. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 22 • Weighted Stouffer’s – Z’s from bigger studies get bigger weights Statistic map E.g. Z-map IBMA on Z maps + N + N (Slide adapted from Thomas Nichols, OHBM 2015)
- 23. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 23 • Random Effects (RFX) GLM – Analyze per-study contrasts as “data” Contrast + standard error maps • Fixed-Effects (FFX) GLM – Don’t estimate variance, just take from first level IBMA on Contrast maps Contrast map Contrast & std. err. maps (Slide adapted from Thomas Nichols, OHBM 2015)
- 24. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 24 Implementations • Not all of these options are easily used Meta-Analysis Method Inputs Neuroimaging Implementation ‘Gold Standard’ MFX Con’s + SE’s FSL’s FEAT SPM spm_mfx AFNI 3dMEMA RFX GLM Stouffer’s RFX Con’s Z’s FSL, SPM, AFNI, etc… FFX GLM Fisher’s Stouffer’s Stouffer’s Weighted Con’s +SE’s Z’s Z’s Z’s + N’s n/a (Slide from Thomas Nichols, OHBM 2015)
- 25. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 25 Self Promotion Alert: IBMA toolbox • SPM Extension • Still in beta! – But welcome all feedback • Available on GitHub https://github.com/NeuroimagingMetaAnalysis/ibma
- 26. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 26 Meta-analysis of 21 pain studies • Results – GLM methods similar – Z-based methods similar • But FFX Z methods more sensitive (as expected) RFX Data: Tracey pain group, FMRIB, Oxford.
- 27. How to publish your statistic maps?
- 28. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 28 Share your statistic maps http://neurovault.org
- 29. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 29 Share your statistic maps http://neurovault.org
- 30. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 30 From SPM & FSL NIDM-Results http://nidm.nidash.org/getting-started/
- 31. • When data available, Image-Based preferred to Coordinate-Based meta-analysis Conclusions Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 31
- 32. • When data available, Image-Based preferred to Coordinate-Based meta-analysis • In practice, it is difficult to use the gold standard Mixed-Effects GLM Conclusions Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 32
- 33. • When data available, Image-Based preferred to Coordinate-Based meta-analysis • In practice, it is difficult to use the gold standard Mixed-Effects GLM • When only contrast estimates are available, RFX GLM is a practical & valid approach Conclusions Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 33
- 34. • When data available, Image-Based preferred to Coordinate-Based meta-analysis • In practice, it is difficult to use the gold standard Mixed-Effects GLM • When only contrast estimates are available, RFX GLM is a practical & valid approach • Few tools for Z-based IBMA, but underway… Conclusions Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 34
- 35. • When data available, Image-Based preferred to Coordinate-Based meta-analysis • In practice, it is difficult to use the gold standard Mixed-Effects GLM • When only contrast estimates are available, RFX GLM is a practical & valid approach • Few tools for Z-based IBMA, but underway… • Data sharing tools: NeuroVault, NIDM-Results Conclusions Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 35
- 36. Camille Maumet - OHBM Neuroimaging Meta-Analysis Educational course 25 June 2017 36 Thank you! This work is supported by