On 10/26/2012 05:10 AM, Andreas Schindler wrote:

Dear freesurfers

Have you ever tried to get an average retinotopic map across all subjects on freesurfer's standard surface (fsaverage)?

I obtained an individual map for each subject using the fsfast stream (mkanalysis-sess, selxavg3-sess) and plotted the results using rtview: e.g.

rtview.sh --s subj1 --lh --polar --real polar/real.nii.gz --imag polar/imag.nii.gz --fsig polar/fsig.nii.gz

Now for the group map, my initial idea was to sample each subject's real and imaginary component to the fsaverage surface and average them separately across subjects. However, I am not sure how I would threshold such an average map on fsaverage using rtview (i.e. what would I use as fsig.nii) given that averaging log p-values is probably not a good idea.

Within each subject's rtopy.self.?/polar(or /eccen) folder the following files (and few more but probably not important here) exist:

fsig.nii -> I understand this contains the unsigned log p values probably obtained via an F-test (frequency of interest) / (noise frequencies) sig.nii -> this might then be the signed counterpart of fsig.nii imag.nii -> this must contain the imaginary component of the polar angle or eccen map real.nii -> this must contain the real component of the polar angle or eccen map mag.nii -> I suppose this might contain the amplitude of the signal at the frequency of interest (i.e. the frequency the wedge / ring is traveling)

However, I don't quite know what the remaining files are good for

I've added info below

minsig.nii -> best p-value between real and imaginary as compared to 0/fixation iminsig.nii -> index of the best p-value (1=real, 2=imag) cwmap-imag.nii -> not useful, for compatability with tksurfer cwmap-real.nii-> not useful, for compatability with tksurfer

ces.nii - two frame file, real and imag cesmag.nii - magnitude cesmagpct.nii - magnitude percent (with respect to voxel mean) cespct.nii - same as ces.nii but percent with respect ot voxel mean cesvar.pct - variance of real, imag (ie, square of the error bars) cesvarpct.nii - variance as a percent of voxel mean^2

For the ces* files I understand that they generally contain effect sizes / variances and are used for group analyses. However, I am not quite sure what effects they correspond to in the case of retinotopy.

When browsing the web for similar problems I found the following post:

...

Hi Michelle, try this:

isxconcat-sess -sf sessidlist -a rtopy.fsaverage.lh -call -o retgroup

cd retgroup/rtopy.fsaverage.lh

mri_glmfit --y eccen/ces.000.nii.gz --osgm --o eccen/glm.real --surface fsaverage lh mri_glmfit --y eccen/ces.001.nii.gz --osgm --o eccen/glm.imag --surface fsaverage lh mri_glmfit --y polar/ces.000.nii.gz --osgm --o polar/glm.real --surface fsaverage lh mri_glmfit --y polar/ces.001.nii.gz --osgm --o polar/glm.imag --surface fsaverage lh

mri_fieldsign --fs lh.fieldsign.mgh \ --eccen eccen/glm.real/osgm/gamma.mgh eccen/glm.imag/osgm/gamma.mgh \ --polar polar/glm.real/osgm/gamma.mgh polar/glm.imag/osgm/gamma.mgh \ --s fsaverage --hemi lh --sphere --old

tksurfer fsaverage lh inflated -aparc -ov lh.fieldsign.mgh -fthresh .5

However, as mentioned above I wonder what effect sizes are used here and why (different to rtview) there is no significance map needed to plot the data.

The above basically does what you suggest above (ie, average the real and imag components across subject). Each of those glmfit commands will produce a sig.mgh file. You can create a single sig file for use with rtview with a conjunction: mri_concat --conjunct eccen/glm.real/osgm/sig.mgh eccen/glm.imag/osgm/sig.mgh polar/glm.real/osgm/sig.mgh polar/glm.imag/osgm/sig.mgh --o conjunction.sig.mgh

You can then load up the relevant files into rtview

doug

Is there a way to use rtview for the result of a group analysis? And if yes how would I get a fsig.nii file for that?

I really appreciate your help!

many thanks& best wishes

Andreas

---

Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer