Freesurfer Group Analysis(FSGD): Difference between revisions

From CCN Wiki
Jump to navigation Jump to search
No edit summary
 
(2 intermediate revisions by the same user not shown)
Line 1: Line 1:
=Freesurfer Group Differences Analysis=
=Freesurfer Group Differences Analysis=
Kinda cool method that can be used to compare groups and/or regress out nuisance variables from BOLD data within Freesurfer.
Kinda cool method that can be used to compare groups and/or regress out nuisance variables from BOLD data within Freesurfer. This method can potentially be used in place of the other GLM group analysis discussed on this wiki using output from selexavg (or prior to since can specify contrasts here in one step).


Some sources:
Some sources:
Line 72: Line 72:
  0.5 0.5
  0.5 0.5


=glm=
=GLM=
Run new GLM for each hemi.
Command:
Command:
  mri_glmfit --surface fsaverage rh --glmdir g2v0 --y T1T2_ces.nii.gz --fsgd g2v0.fsgd --C GroupDiff.mtx --C GroupT1.mtx --C GroupT2.mtx --C GroupT1T2.intercept.mtx  
  mri_glmfit --surface fsaverage rh --glmdir g2v0 --y T1T2_ces.nii.gz --fsgd g2v0.fsgd --C GroupDiff.mtx --C GroupT1.mtx --C GroupT2.mtx --C GroupT1T2.intercept.mtx


=Correct for multi comparisons=
=Correct for Multiple Comparisons=
Next:
Next:
  mri_glmfit-sim --glmdir g2v0 --cache 3 abs --cwpvalthresh .05
  mri_glmfit-sim --glmdir g2v0 --cache 3 abs --cwpvalthresh .05

Latest revision as of 17:36, 30 May 2018

Freesurfer Group Differences Analysis

Kinda cool method that can be used to compare groups and/or regress out nuisance variables from BOLD data within Freesurfer. This method can potentially be used in place of the other GLM group analysis discussed on this wiki using output from selexavg (or prior to since can specify contrasts here in one step).

Some sources:

https://surfer.nmr.mgh.harvard.edu/fswiki/FsgdExamples
https://surfer.nmr.mgh.harvard.edu/fswiki/FsgdFormat
https://surfer.nmr.mgh.harvard.edu/fswiki/Fsgdf2G0V


Data

If comparing T1 and T2 data, grab the files used for original GLMs (e.g. ces.nii.gz). Rename and cat them for each hemi:

mri_concat T1_ces.nii.gz T2_ces.nii.gz --o T1T2_ces.nii.gz

FSGD File

For T1 vs. T2 booth data analysis, for example, with 2 groups and no covariates (g2v0.fsgd).

The first line needs to stay the same 'GroupDescriptorFile 1'.

GroupDescriptorFile 1
Title OSGM_T1T2
Class GroupT1
Class GroupT2
Input FS_T1_501 GroupT1
Input FS_T1_505 GroupT1
Input FS_T1_506 GroupT1
Input FS_T1_508 GroupT1
Input FS_T1_513 GroupT1
Input FS_T1_515 GroupT1
Input FS_T1_529 GroupT1
Input FS_T1_531 GroupT1
Input FS_T1_534 GroupT1
Input FS_T1_542 GroupT1
Input FS_T1_546 GroupT1
Input FS_T1_557 GroupT1
Input FS_T1_564 GroupT1
Input FS_T1_575 GroupT1
Input FS_T1_576 GroupT1
Input FS_T1_579 GroupT1
Input FS_T1_580 GroupT1
Input FS_T1_587 GroupT1
Input FS_T1_589 GroupT1
Input FS_T2_501 GroupT2
Input FS_T2_505 GroupT2
Input FS_T2_506 GroupT2
Input FS_T2_508 GroupT2
Input FS_T2_513 GroupT2
Input FS_T2_515 GroupT2
Input FS_T2_529 GroupT2
Input FS_T2_531 GroupT2
Input FS_T2_534 GroupT2
Input FS_T2_542 GroupT2
Input FS_T2_546 GroupT2
Input FS_T2_557 GroupT2
Input FS_T2_564 GroupT2
Input FS_T2_575 GroupT2
Input FS_T2_576 GroupT2
Input FS_T2_579 GroupT2
Input FS_T2_580 GroupT2
Input FS_T2_587 GroupT2
Input FS_T2_589 GroupT2

'DODS' contrast files

groupdiff.mtx

1 -1

T1.mtx

1 0

T2.mtx

0 1

T1T2intercept.mtx

0.5 0.5

GLM

Run new GLM for each hemi. Command:

mri_glmfit --surface fsaverage rh --glmdir g2v0 --y T1T2_ces.nii.gz --fsgd g2v0.fsgd --C GroupDiff.mtx --C GroupT1.mtx --C GroupT2.mtx --C GroupT1T2.intercept.mtx

Correct for Multiple Comparisons

Next:

mri_glmfit-sim --glmdir g2v0 --cache 3 abs --cwpvalthresh .05


Now you've got clusters and .annot files to load up and get sizes/means etc.

Overlay

tksurfer fsaverage lh pial

load overlay -> cache.th30.abs.sig.cluster.mgh

cache.th30.abs.sig.cluster.summary contains cluster info.

Alternatively, you can segment the cache.th30.abs.sig.cluster.annot file into desired clusters.