1

Integrating FreeSurfer and FSL FEAT

2

Outline

• Registering FEAT FreeSurfer Anatomical• Automatic (FLIRT)• Manual (tkregister2)

• Surface-based Group fMRI Analysis • Individual Analysis

•Viewing FEAT output on Anatomical• Sampling FEAT output on the surface• Mapping FreeSurfer Segmentations to FEAT

3

RegistrationFreeSurfer Anatomical (orig) Template Functional

Note: Registering the template functional volume to the anatomical volume is sufficientto register the template to the surface.

4

FreeSurfer Registration

FreeSurferSubject-Specific

• Volumes• Surfaces• Thickness• ROIs

Your Data/Software• fMRI (FSL, etc)• EEG/MEG• DTI• …

Registration

Registration Matrix• Affine 4x4• As many as 12 DOF (usually 6)• Text file

5

Automatic Registration

reg-feat2anat –featdir fbert1.feat –subject bert

Uses BBR to perform 6DOF registration: fbert1.feat/example_func.nii.gz $SUBJECTS_DIR/bert/mri/brainmask.mgz

Creates FreeSurfer registration matrix: fbert1.feat/reg/freesurfer/anat2exf.register.dat

reg-feat2anat --help

First: analyze your data with FEAT

Also creates: fbert1.feat/reg/freesurfer/anat2std.register.datThis matrix maps standard space to anatomical space and can be used when combining data within subject across data sets

6

Manual Registration

tkregister2 --help

• reg-feat2anat –feat fbert.feat --manual•Visually inspect registration• Manually edit registration (6 DOF)• Cf Manual Talairach registration

7

Tips• Rigid = 6 DOF = No stretching

• Use CSF to get a sense of where the folds are

• Avoid using B0 distortion regions

• Avoid using ventricles

• Warning about “edge” of the brain

• Same Subject, Left-Right Flips

8

Sampling on the Surface

White/GrayPial

• White/Gray• Pial• Half Way• Average

Projection Fraction

--projfrac 0.5

9

Sampling on the Surface

10

Surface-based fMRI Group Analysis

11

Surface-based fMRI Group Analysis

S1R1

S2R2

S3R3

Subject 1

Subject 2

Subject 3

FunctionalCOPEs

fsaverage

Can be com

paredV

oxel-for-voxel}… reg-feat2anat recon-all

mris_preproc

12

Surface-based fMRI Group Analysis (One Run)

Subject1

run1.feat

stats

cope1.nii

• Single Run •Analyze with FEAT, No smoothing• COPEs are in native functional space

13

Surface-based fMRI Group Analysis (One Run)

Subject1

run1.feat

stats

cope1.nii

• run reg-feat2anat

reg

freesurfer

anat2exf.register.dat – anatomical–example_func registration

reg-feat2anat

14

Surface-based Group fMRI Analysis (One Run)mris_preproc --out lh.cope1.nii.gz --target fsaverage --hemi lh --iv bert.feat/stats/cope1.nii.gz bert.feat/reg/freesurfer/anat2exf.register.dat --iv greg.feat/stats/cope1.nii.gz greg.feat/reg/freesurfer/anat2exf.register.dat --iv sally.feat/stats/cope1.nii.gz sally.feat/reg/freesurfer/anat2exf.register.dat --iv pat.feat/stats/cope1.nii.gz pat.feat/reg/freesurfer/anat2exf.register.dat

Volumes Registrations

• lh.cope1.mgh – stack of subjects• Can map any functional data, eg, zstat, fzstat, cope, pe, etc• fsaverage – defines common space (spherical surface)• Group analysis same as with a thickness study:

• surface smoothing, mri_glmfit, mri_glmfi-sim

15

Surface-based fMRI Group Analysis (>1 Run)

Subject1

run1.feat

stats

cope1.nii

run2.feat

stats

cope1.nii

• Multiple Runs •Analyze each run with FEAT, No smoothing• COPEs are in native functional space

16

Subject1

run1.feat

stats

cope1.nii

run2.feat

stats

cope1.nii

• Merge runs with GFEAT with Fixed Effects• GFEAT results are in MNI152 Standard Space

xrun.gfeat

cope1.feat

mean_func.nii stats

cope1.nii

MNI152

Surface-based fMRI Group Analysis (>1 Run)

17

Subject1

run1.feat run2.feat xrun.gfeat

reg

freesurfer

anat2exf.register.dat – anatomical–example_func registrationanat2std.register.dat -- anatomical–standard MNI152 space registration

stats

cope1.nii

reg-feat2anat

• run reg-feat2anat on one run• anat2std.register.dat – same across all runs

cope1.feat

mean_func.nii stats

cope1.nii

MNI152

Surface-based fMRI Group Analysis (>1 Run)

18

Subject1

run1.feat run2.feat xrun.gfeat

reg

freesurfer

anat2exf.register.dat – anatomical–example_func registrationanat2std.register.dat -- anatomical–standard MNI152 space registration

stats

cope1.nii

reg-feat2anat

cope1.feat

mean_func.nii stats

cope1.nii

MNI152

Verify the registrationtkregister2 –mov xrun.gfeat/cope1/mean_func.nii.gz --reg run1.feat/reg/freesurfer/anat2std.register.dat --surf

Surface-based fMRI Group Analysis (>1 Run)

19

mris_preproc --out lh.cope1.mgh --target fsaverage --hemi lh --iv bert.gfeat/cope1.feat/stats/cope1.nii.gz bert.feat/reg/freesurfer/anat2std.register.dat --iv greg.gfeat/cope1.feat/stats/cope1.nii.gz greg.feat/reg/freesurfer/anat2std.register.dat --iv sally.feat/cope1.feat/stats/cope1.nii.gz sally.feat/reg/freesurfer/anat2std.register.dat --iv pat.feat/cope1.feat/stats/cope1.nii.gz pat.feat/reg/freesurfer/anat2std.register.dat

Volumes Registrations

Surface-based fMRI Group Analysis (>1 Run)

• lh.cope1.mgh – stack of subjects• Can map any functional data, eg, zstat, fzstat, cope, pe, etc• fsaverage – defines common space (spherical surface)• Group analysis same as with a thickness study:

• surface smoothing, mri_glmfit, mri_glmfi-sim

20

Surface-based fMRI Group AnalysisLeft Hemi Right Hemi

fBIRN Group n=18, working memory, distractor-vs-fix

Applying FreeSurfer Tools to FSL fMRI Analysis 21

Individual Subject Integration

22

Viewing Functional Activation on the Volume

tkmedit bert orig.mgz -aux brain.mgz -overlay ./fbert1.feat/stats/zstat1.nii.gz -overlay-reg ./fbert1.feat/reg/freesurfer/anat2exf.register.dat -fthresh 2.3 –fmax 4.3 -seg aparc+aseg.mgz

Thresholds depend on the nature of the data, Eg, for zstat image, 2.3 means z > 2.3 Can be changed with View->Configure->FunctionalOverlay

Visual, Auditory, Motor

Can display any functional data, eg,zstat, fzstat, cope, pe, etc

23

Volume Viewing

• Red/Yellow +• Blue/Cyan -• Seg Opacity• ROI Average• ROI Count

24

Viewing FEAT Stats on the Surfacetksurfer bert rh inflated -overlay ./fbert1.feat/stats/zstat1.nii.gz -overlay-reg ./fbert1.feat/reg/freesurfer/anat2exf.register.dat -fthresh 2.3 -fmid 3.3 -fslope 1

Visual, Auditory, Motor

Can display any functional data, eg,zstat, fzstat, cope, pe, etc

25

Surface Viewing

• Red/Yellow +,Blue/Cyan -• Parcellation Outline• ROI Average• ROI Count

26

Mapping Automatic Segmentations

$FREESURFER_HOME/FreeSurferColorsLUT.txtMaps numerical index to ROI name, eg Left-Hippocamus = 17

aparc+aseg.mgzlh.aparc.annotaseg.mgz

27

Mapping Automatic Segmentations

aseg2feat --feat fbert.feat –aseg aparc+aseg

In the functional FOV, creates:fbert1.feat/reg/freesurfer/aseg+aparc.nii.gz

Create Text Summary Table (nvox, mean cope, std cope, etc)mri_segstats --seg fbert1.feat/reg/freesurfer/aparc+aseg.nii.gz --nonempty --ctab-default --in fbert.feat/stats/cope1.nii.gz --sum fbert1.sum.txt

Can summarize any functional data, eg,zstat, fzstat, cope, pe, etc

28

Make and View ROI

tkmedit bert orig.mgz -aux brain.mgz -overlay ./fbert1.feat/reg/freesurfer/lh.putamen.nii.gz -overlay-reg ./fbert1.feat/reg/freesurfer/anat2exf.register.dat -fthresh 0.5 -seg aparc+aseg.mgz

Make a binary mask of the left putamen:Note: 12 = Left Putamen, see $FREESURFER_HOME/FreeSurferColorsLUT.txt

fslmaths ./fbert1.feat/reg/freesurfer/aparc+aseg.nii.gz -thr 12 -uthr 12 ./fbert1.feat/reg/freesurfer/lh.putamen.nii.gz

29

Practical Data• Sensory-motor study • Blocked Design (15sec OFF, 15 sec ON)• One subject – “bert”• Two runs (TR=3, 85 time points)• FEAT has been run on both runs (FWHM=5)• Combined with GFEAT

– FFx– One-Sample Group Mean (OSGM)

• Actually, all analysis steps already performed!

30

Practical• Automatic registration (<5 min)• Manual registration• View FEAT results on subject’s anatomy (aparc+aseg)• View FEAT results with tksurfer• Map aparc+aseg to Functional Space• Verify GFEAT registration• View GFEAT results in volume and on surface• “Higher-Level” analysis with mri_glmfit

– Cross-Run

– Fixed-Effects (FFX), One-Sample Group Mean (osgm)

31

End of Presentation

32

-0.1

+0.5+0.3

+0.10.0 (white)

+1.0 (pial)+0.9

+0.7

+1.1

Sampling on the Surface: Projection Fraction

33

Native Anatomical Surface Space

Step 1: Register Anatomical with Surface Atlas (fsaverage)

S

recon-all

Anatomical Surface in fsaverage Space

fsaverage SpaceSurface-basedRegistration

34

Native Anatomical SpaceNative Functional Space

R

RigidRegistration

reg-feat2anatbbregister

Step 2: Register fMRI with Anatomical

fMRI inAnatomical Space

35

Native Anatomical Space

Step 3: Combine Steps 1 and 2: mris_preproc

S

recon-all

fMRI infsaverage Space

fsaverage SpaceNative Functional Space

R

reg-feat2anatbbregister

36

Within-Subject, Cross-Run Analysis

• Analyze each run with FEAT (dataX.feat)• Combine runs with GFEAT (standard space)• mean_func.nii.gz – avg of example_func in std space• Register each run (not .gfeat) with reg-feat2anat. • dataX.feat/reg/freesurfer/anat2std.register.dat• All runs (X) should be very close

Verify the registrationtkregister2 –mov data.gfeat/mean_func.nii.gz --reg anat2std.register.dat --surf

Use anat2std.register.dat the way you would anat2exf.register.dat