Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
University of Warwick, Warwick Manufacturing Group & Department of Statistics, Coventry, UK.
Camille Maumet and Thomas E. ...
Agenda
•  Meta-analyses in Neuroimaging
–  Why?
–  Coordinate-Based or Image-Based?
•  Image-Based Meta-Analysis
–  Gold s...
Meta-Analyses in Neuroimaging
3
Why meta-analyses?
•  Power increase
•  Combine information across studies
Data
acquisition
Analysis
Experiment Raw data R...
Data analysis in neuroimaging
Analysis
ResultsExperiment
Data
acquisition
Raw data Paper
Publication
Imaging
data
5
500MB/...
Coordinate-Based Meta-Analysis
Table of local maxima
(quantitative)
6
Paper
Publication
?
Detection images
(qualitative)
P...
Coordinate- or Image-Based?
7
Data
acquisition
Analysis
Experiment Raw data Results
Data
acquisition
Analysis
Experiment R...
Image-Based Meta-Analysis
8
Meta-analysis levelStudy levelSubject level
Meta-analysis gold standard
Pre-processed
data
Subject1
Model fitting
and esti...
Image-based Meta-analysis
•  Gold standard:
•  But…
–  Units will depend on:
•  The scaling of the data (subject-level)
• ...
Image-Based Meta-Analysis
•  Other (sub-optimal) statistics available:
–  Based on z-statistic:
•  Fishers’s; Stouffer’s; ...
IBMA toolbox
•  Plugin for
•  Available on github:
https://github.com/NeuroimagingMetaAnalysis/ibma
12
Validity of IBMA approaches in
neuroimaging
13
Meta-analysis of 21 pain studies
•  Data
–  21 studies investigated pain in healthy subjects
14
Conclusion
•  Towards Image-Based meta-analysis.
•  In practice, it is difficult to use the gold
standard Third-level Mixe...
Acknowledgements
Q & A
We gratefully acknowledge
the use of MRI data from
the Tracey pain group,
FMRIB, Oxford. 	

	

This...
Upcoming SlideShare
Loading in …5
×

IBMA: An SPM toolbox for Neuroimaging Image-Based Meta-Analysis

546 views

Published on

While most neuroimaging meta-analyses are based on peak coordinate data, the best practice method is an image-based meta-analysis that combines the effect estimates and the standard errors from each study [7]. Various efforts are underway to facilitate sharing of neuroimaging data to make such intensity-based meta-analysis possible (see, e.g. [4]).
When image data is available for each study, a number of approaches (see [6] for a review) have been proposed to perform such meta-analysis including combination of standardised statistics, just effect estimates or both effects estimates and their sampling variance. While the latter is the preferred approach in the statistical community [1], often only standardised estimates are shared, reducing the possible meta-analytic approaches.
In view of the increasing availability of image data for neuroimaging analyses, we introduce IBMA, a toolbox for SPM [8] providing a set of tools for image-based meta-analysis. The toolbox is freely available at: https://github.com/NeuroimagingMetaAnalysis/ibma.

Published in: Science
  • Be the first to comment

  • Be the first to like this

IBMA: An SPM toolbox for Neuroimaging Image-Based Meta-Analysis

  1. 1. University of Warwick, Warwick Manufacturing Group & Department of Statistics, Coventry, UK. Camille Maumet and Thomas E. Nichols IBMA: An SPM toolbox for Neuroimaging Image-Based Meta-Analysis
  2. 2. Agenda •  Meta-analyses in Neuroimaging –  Why? –  Coordinate-Based or Image-Based? •  Image-Based Meta-Analysis –  Gold standard –  Other approaches •  Validity of IBMA approaches in neuroimaging 2
  3. 3. Meta-Analyses in Neuroimaging 3
  4. 4. Why meta-analyses? •  Power increase •  Combine information across studies Data acquisition Analysis Experiment Raw data Results Data acquisition Analysis Experiment Raw data Results … Results Meta- analysis 4
  5. 5. Data analysis in neuroimaging Analysis ResultsExperiment Data acquisition Raw data Paper Publication Imaging data 5 500MB/subject 20GB 2.5GB/subject 100GB [ ~2GB for stats] < 0.5MB0MB
  6. 6. Coordinate-Based Meta-Analysis Table of local maxima (quantitative) 6 Paper Publication ? Detection images (qualitative) Peaks (quantitative) < 0.5MB
  7. 7. Coordinate- or Image-Based? 7 Data acquisition Analysis Experiment Raw data Results Data acquisition Analysis Experiment Raw data Results … Publication Publication Paper Paper Coordinate-based meta-analysis Image-based meta-analysis Shared results Data sharing
  8. 8. Image-Based Meta-Analysis 8
  9. 9. Meta-analysis levelStudy levelSubject level Meta-analysis gold standard Pre-processed data Subject1 Model fitting and estimation Contrast and std. err. maps Inference Detections (subject-level) Model fitting and estimationPre-processed data Subjectn Contrast and std. err. maps … Inference Detections (subject-level) Model fitting and estimation Contrast and std. err. maps Inference Detections (study-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 Model fitting and estimation Contrast and std. err. maps Inference Detections (meta-analysis) Inference Detections (study-level) 9
  10. 10. Image-based Meta-analysis •  Gold standard: •  But… –  Units will depend on: •  The scaling of the data (subject-level) •  The scaling of the predictor(s) that are involved in the selected contrast (subject- and study-level) •  The scaling of the selected contrast (subject- and study- level). –  Contrast estimates and standard error maps are rarely shared… 10 Third-level Mixed-Effects GLM
  11. 11. Image-Based Meta-Analysis •  Other (sub-optimal) statistics available: –  Based on z-statistic: •  Fishers’s; Stouffer’s; “Stouffers’s MFX” –  Based on z-statistic + sample size •  Weighted Stouffer’s –  Based on contrast estimates only: •  RFX GLM; –  Based on contrast estimates and standard error •  Fixed-Effects GLM •  Based on restrictive assumptions, robustness to violation need to be further studied 11
  12. 12. IBMA toolbox •  Plugin for •  Available on github: https://github.com/NeuroimagingMetaAnalysis/ibma 12
  13. 13. Validity of IBMA approaches in neuroimaging 13
  14. 14. Meta-analysis of 21 pain studies •  Data –  21 studies investigated pain in healthy subjects 14
  15. 15. Conclusion •  Towards Image-Based meta-analysis. •  In practice, it is difficult to use the gold standard Third-level Mixed-Effects General Linear Model. •  IBMA toolbox provides alternative approaches. •  Further investigation: two-sample analysis… 15
  16. 16. Acknowledgements Q & A We gratefully acknowledge the use of MRI data from the Tracey pain group, FMRIB, Oxford. This work is supported by the 16

×