meta-analysis practical intensity-based · practical intensity-based meta-analysis camille maumet...
TRANSCRIPT
Practical Intensity-BasedMeta-Analysis
Camille MaumetOHBM Neuroimaging Meta-Analysis Educational course
26 June 2016
Coordinate- or Image-Based?
2
Acquisition Analysis
Experiment Raw data Results
Acquisition Analysis
Experiment Raw data Results
…
Publication
Publication
Paper
Paper
Coordinate- or Image-Based?
2
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
Shared resultsData sharing
Coordinate- or Image-Based?
2
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
Image-based meta-analysishow to?
3
InferenceDetections
(subject-level)
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
4
InferenceDetections
(subject-level)
InferenceDetections
(subject-level)
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
…
4
InferenceDetections
(study-level)
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
… Model fitting and estimation Contrast and
std. err. maps
4
InferenceDetections
(study-level)
InferenceDetections
(study-level)
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
… Model fitting and estimation Contrast and
std. err. maps
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
… Model fitting and estimation Contrast and
std. err. maps
4
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
… Model fitting and estimation Contrast and
std. err. maps
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
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)
4
InferenceDetections
(subject-level)
InferenceDetections
(subject-level)
InferenceDetections
(study-level)
InferenceDetections
(study-level)
Meta-analysis levelStudy levelSubject level
Image-based meta-analysis
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
Contrast and std. err. maps
… Model fitting and estimation Contrast and
std. err. maps
Pre-processed dataS
ubje
ct 1
Model fitting and estimation Contrast and
std. err. maps
Model fitting and estimationPre-processed
dataSub
ject
n
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)
4
Image-based meta-analysis• Gold standard: Third-level Mixed-Effects GLM• Requirements
– study-level Contrast estimates and Standard error maps.
– Same units
Contrast and std. err. maps
5
Units of contrast estimatesPre-processed
data
Model fitting and estimation Contrast and
std. err. maps
6
Units of contrast estimatesPre-processed
data
Model fitting and estimation Contrast and
std. err. maps
6
Pre-processed data
Data scalingScaled pre-proc. data
Model parameter estimation Parameter
estimates
Contrast estimation Contrast and
std. err. maps
Units depend on mean estimation and scaling target.
Units of contrast estimatesPre-processed
data
Data scalingScaled pre-proc. data
Model parameter estimation Parameter
estimates
Contrast estimation Contrast and
std. err. maps
Data scaling
7
Units of contrast estimates
Y = β +
Units depend on scaling of explanatory variables
Pre-processed data
Data scalingScaled pre-proc. data
Model parameter estimation Parameter
estimates
Contrast estimation Contrast and
std. err. maps
Model parameter estimation
8
Units of contrast estimates
• 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 scalingScaled pre-proc. data
Model parameter estimation Parameter
estimates
Contrast estimation Contrast and
std. err. maps
9
Contrast estimation
Image-based Meta-analysis• 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…
10
Third-level Mixed-Effects GLM
3dMEMA_result+tlrc.BRIK[[0]][from contrast & stat maps]
Which images for IBMA?
Contrast & std. err. maps
Statistic mapE.g. Z-map
Contrast map
SPM FSL AFNI
con_0001.nii[SPM.mat]
cope1.niivarcope1.nii (squared)
3dMEMA_result+tlrc.BRIK[[1]]spmT_0001.nii tstat1.nii.gzzstat1.nii.gz
3dMEMA_result+tlrc.BRIK[[0]]con_0001.nii cope1.nii
11
Image-based meta-analyses based on Z
• 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
12(Slide adapted from Thomas Nichols, OHBM 2015)
Statistic mapE.g. Z-map
• Weighted Stouffer’s
– Z’s from bigger studies get bigger weight
13(Slide adapted from Thomas Nichols, OHBM 2015)
Statistic mapE.g. Z-map
Image-based meta-analyses based on Z + N + N
• Random Effects (RFX) GLM
– Analyze per-study contrasts as “data”
based on Con’s + SE’s• Fixed-Effects (FFX) GLM
– Don’t estimate variance, just take from first level
14(Slide adapted from Thomas Nichols, OHBM 2015)
Image-based meta-analyses based on Con’s
Contrast map
Contrast & std. err. maps
Image-Based Meta-AnalysisIn practice!• Not all of these options are easily used
15
Meta-Analysis Method Inputs Neuroimaging Implementation
‘Gold Standard’ MFX Con’s + SE’s FSL’s FEATSPM spm_mfxAFNI 3dMEMA
RFX GLMStouffer’s RFX
Con’sZ’s
FSL, SPM, AFNI, etc…
FFX GLMFisher’sStouffer’sStouffer’s Weighted
Con’s +SE’sZ’sZ’sZ’s + N’s
n/a
(Slide from Thomas Nichols, OHBM 2015)
Self Promotion Alert:IBMA toolbox• SPM Extension• Still in beta!
– But welcome all feedback
• Available on GitHub https://github.com/NeuroimagingMetaAnalysis/ibma
16
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. 17
Share image data supporting neuroimaging results
From SPM & FSL
NIDM-Results
http://nidm.nidash.org/getting-started/21
• When data available, Image-Based preferred to Coordinate-Based meta-analysis
Conclusions
For more on NIDM-ResultsMaumet et al., Poster 1851 - Tuesday 12:45-14:45“NIDM-Results: Standardized reporting of mass univariate neuroimaging results in SPM, FSL and AFNI”
22
Conclusions• When data available, Image-Based preferred to
Coordinate-Based meta-analysis• In practice, it is difficult to use the gold standard
Mixed-Effects GLM
For more on NIDM-ResultsMaumet et al., Poster 1851 - Tuesday 12:45-14:45“NIDM-Results: Standardized reporting of mass univariate neuroimaging results in SPM, FSL and AFNI”
22
Conclusions• 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
For more on NIDM-ResultsMaumet et al., Poster 1851 - Tuesday 12:45-14:45“NIDM-Results: Standardized reporting of mass univariate neuroimaging results in SPM, FSL and AFNI”
22
Conclusions• 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…
For more on NIDM-ResultsMaumet et al., Poster 1851 - Tuesday 12:45-14:45“NIDM-Results: Standardized reporting of mass univariate neuroimaging results in SPM, FSL and AFNI”
22
Conclusions• 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
For more on NIDM-ResultsMaumet et al., Poster 1851 - Tuesday 12:45-14:45“NIDM-Results: Standardized reporting of mass univariate neuroimaging results in SPM, FSL and AFNI”
22
Thank you!
This work is supported by the