It looks like you have it correct. The docs are admittedly incomplete, but they are slowly getting updated. One of the things that is changing is that I'm transitioning away from the isxavg programs to mri_glmfit, which is much more flexible for doing random group and inter-group analysis. mri_glmfit can be used in conjunction with mri_volcluster or mri_surfcluster to perform clustering/correction for mult comparisons. Setting thresholds for clustering is not done as part of the analysis. Rather, it is done at the end as you may want to change these parameters.
The new work-flow will be: 1. Preprocessing, registration 2. Analysis with selxavg and stxgrinder 3. Resampling to common space (vol or surf) and concatenating 4. Group analysis with mri_glmfit 5. Clustering with mri_volcluster or mri_surfcluster
The one piece that I am refining is step 3. The surface part can be done with mris_preproc, though I've written a new script called isxconcat-sess which will do both the volume and the surface. I believe this works now, but I have not tested it much.
There are more docs on using mri_glmfit and the clustering on the wiki surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferGroupAnalysis. This is geared towards doing surface-based thickness group studies, but it applies to volume-based as well as fmri data.
Sorry, I know the details on this are a little sketchy. I'll fill out the wiki more when I'm done.
doug
Avram Holmes wrote:
Dear Doug,
Thanks for your reply. As new users, we're having a little trouble verifying that our processing steps are correct. We're really after two things. with a third question regarding support materials:
- Some feedback regarding whether we've completed all the steps
necessary for a glm analysis. Based on our reading and (very limited) experience, it seems like the combination of the mkanalysis, mkcontrast, selxavg, and stxgrinder steps are all we need to do this properly, and that using isxavg will allow us to run a second-level group analysis. Is this correct? We're concerned that we may have missed something along the way.
- Insight into how to set cluster extents and conduct statistical
tests on the data. We're used to SPM and are struggling a bit with the inferential side of FsFast.
- We have been passed along the Handbook for the MGH-NMR standard
processing stream (which was put out in 2000). Do you know if there is a more current resource? The tutorial on the Wiki is a little incomplete and I haven't been able to chase down anything on how to run a glm in FsFast.
Any help will be greatly appreciated.
Thanks!
Avram
Avram Holmes Department of Psychology Harvard University 1210 William James Hall Phone:(617) 495-0790 33 Kirkland Street Fax:(617) 495-3728 Cambridge, MA 02138, USA Email: holmes@fas.harvard.eduOn Tue, 12 Dec 2006, Doug Greve wrote:
Just looking at your scripts, they look ok. It's hard to figure out what "trouble viewing/interpreting the results" means.
doug
Avram Holmes wrote:
FSFast support folks,
I am looking for some support running an analysis in FSFast. As far as we can tell we have successfully run every step in a random effects analysis but we are having trouble viewing/interpreting the results. I was hoping that I would be able to post my processing stream and see if anyone would give us feedback on what we have or haven?t done correctly (see below). My understanding is that after I run mkanalysis-sess.new, selxavg-sess, mkcontrast-sess, stxgrinder-sess, and isxavg-re-sess, I have fit a model to the data and I am testing the fit of the model across our contrasts. However, I?m not seeing the expected effects and I am wondering if I am missing a step or if my stream is not optimized. I have spent a considerable amount of time searching the wiki page and I haven?t been able to track down a detailed description of how to conduct and interpret a complete GLM analysis in FSFast so I am a little unsure if I have everything correct.
Any help would be greatly appreciated, Avram
#! /bin/csh -f
Step1
set mypath = . set sub = $1 mc-sess -fstem f -fmcstem fmc -s ${sub} -d ${mypath} stc-sess -i fmc -o fmcstc -so siemens -s ${sub} -d ${mypath} spatialsmooth-sess -i fmcstc -o fmcstcsm5 -fwhm 5 -d ${mypath} -s ${sub} mkbrainmask-sess -s ${sub} -d ${mypath}
Step2
set analysis = 'moneygamma_test' set subj = 'FPPREW01 FPPREW02 FPPREW03 FPPREW04 FPPREW05 FPPREW06 FPPREW07 FPPREW08 FPPREW09 FPPREW10 FPPREW11 FPPREW12 FPPREW13 FPPREW14 FPPREW15 FPPREW16 FPPREW17 FPPREW18'
mkanalysis-sess.new -analysis ${analysis} -gammafit 2.25 1.25 -TR 2.5 -paradigm moneygame.par -designtype event-related -timewindow 30 -funcstem fmcstcsm5 -nconditions 5 -polyfit 1 -mcextreg -taumax 20 -mask brain
foreach s ($subj) selxavg-sess -s ${s} -d . -analysis ${analysis} end
mkcontrast-sess -analysis ${analysis} -contrast omnibus -a 1 -a 2 -a 3 -a 4 -a 5 -c 0 -nosumconds mkcontrast-sess -analysis ${analysis} -contrast pun_v_noinc -a 1 -c 2 mkcontrast-sess -analysis ${analysis} -contrast rew_v_noinc -a 3 -c 2 mkcontrast-sess -analysis ${analysis} -contrast pun_v_rew -a 1 -c 3 mkcontrast-sess -analysis ${analysis} -contrast rew_v_pun -a 3 -c 1
foreach s ($subj) autoreg-sess -d . -s ${s} -fsd bold end
Step3
set mypath = . set analysis = 'moneygamma_test' set subj = 'FPPREW01 FPPREW02 FPPREW03 FPPREW04 FPPREW05 FPPREW06 FPPREW07 FPPREW08 FPPREW09 FPPREW10 FPPREW11 FPPREW12 FPPREW13 FPPREW14 FPPREW15 FPPREW16 FPPREW17 FPPREW18'
foreach s ($subj) stxgrinder-sess -contrast omnibus -analysis ${analysis} -s ${s} -d ${mypath} stxgrinder-sess -contrast pun_v_noinc -analysis ${analysis} -s ${s} -d ${mypath} stxgrinder-sess -contrast rew_v_noinc -analysis ${analysis} -s ${s} -d ${mypath} stxgrinder-sess -contrast pun_v_rew -analysis ${analysis} -s ${s} -d ${mypath} stxgrinder-sess -contrast rew_v_pun -analysis ${analysis} -s ${s} -d ${mypath} end
Step4
set analysis = 'moneygamma_test' set subj = 'FPPREW01 FPPREW02 FPPREW03 FPPREW04 FPPREW05 FPPREW06 FPPREW07 FPPREW08 FPPREW09 FPPREW10 FPPREW11 FPPREW12 FPPREW13 FPPREW14 FPPREW15 FPPREW16 FPPREW17 FPPREW18' set glist = 'GROUPLIST' set group = 'Controls' set contrasts = 'omnibus pun_v_noinc rew_v_noinc pun_v_rew rew_v_pun'
foreach s ($subj) foreach anal ($analysis) func2tal-sess -res 4 -analysis ${anal} -d . -s ${s} foreach hemi (lh rh) func2sph-sess -analysis ${anal} -hemi ${hemi} -projfrac .3 -d . -s ${s} sphsmooth-sess -smoothsteps 10 -analysis ${anal} -insphdir sph -outsphdir sphsm10 -hemi ${hemi} -d . -s ${s} end end end
foreach c ($contrasts) isxavg-re-sess -analysis ${analysis} -group ${group} -space tal -d . -sf ${glist} -contrast ${c} foreach hemi (lh rh) isxavg-re-sess -analysis ${analysis} -group ${group} -space sph -hemi ${hemi} -d . -sf ${glist} -contrast ${c} end end
Avram Holmes Department of Psychology Harvard University 1210 William James Hall Phone:(617) 495-0790 33 Kirkland Street Fax:(617) 495-3728 Cambridge, MA 02138, USA Email: holmes@fas.harvard.edu
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
In order to help us help you, please follow the steps in: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer