Hi Doug,

Thanks for taking a look at the commands!

I would like to ask a further question about your concern/s regarding this approach? My aim was to have an analytical pipeline in Freesurfer that is analogous to several studies (using VBM-type of analyses in SPM) where they compared the relative severity of various pathological processes measured from different modalities.

I have attached a little screenshot for your reference.

For example, the authors in that study were trying to see whether hypometabolism was relatively more severe compared to atrophy, and both modalites (PET and T1) were Z-scored relative to controls, and then compared against each other. Another study that has used a similar technique is here: https://www.ncbi.nlm.nih.gov/pubmed/24951642

Best Wishes, Elijah

Re: [Freesurfer] Comparing z score maps between 2 modalities in a group? http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=subject:%22Re%5C%3A+%5C%5BFreesurfer%5C%5D+Comparing+z+score+maps+between+2+modalities+in+a+group%5C%3F%22&o=newest

Douglas N Greve http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=from:%22Douglas+N+Greve%22 Fri, 18 Nov 2016 08:19:47 -0800 http://www.mail-archive.com/search?l=freesurfer@nmr.mgh.harvard.edu&q=date:20161118

Those commands look right, though I would do the smoothing before computing the standard dev. Also, I don't know whether your overall method is valid.

On 11/15/2016 05:48 PM, Elijah Mak wrote:

Hi Freesurfer Team,

We would like to compare Z-scored surface maps between different modalities (MRI and PET) within a single group. 1 subject = 2 measures.

These are my steps so far, using mri_concat and fscalc for the generation of the Z score maps. Differences between Z-score maps are tested using the paired t-test approach as outlined here: https://surfer.nmr.mgh.harvard.edu/fswiki/PairedAnalysis

I am not sure if my following steps are correct, so confirmation will be much appreciated, many thanks for your time and help!

// Generate mean/sd of healthy controls

X mri_concat --f stacks_X_lh_HC_fsaverage --o group_mean_X_hc_lh.mgh --mean mri_concat --f stacks_X_lh_HC_fsaverage --o group_std_X_hc_lh.mgh --std

Y mri_concat --f stacks__Y_HC_fsaverage --o group_mean_Y_HC_lh.mgh --mean mri_concat --f stacks_Y_lh_HC_fsaverage --o group_std_Y_HC_lh.mgh --std

QN: Should smoothing be applied here or before?

// Derive individual Z-score map for X and Y

X for i in `cat list`; do fscalc $i/surf/lh.X.fsaverage.mgh sub group_mean_X_hc_lh.mgh div group_std_X_hc_lh.mgh --o $i/surf/lh.Z.thickness.fsaverage.mgh; done for i in `cat list; do fscalc $i/surf/rh.X.fsaverage.mgh sub group_mean_cth_hc_rh.mgh div group_std_X_hc_rh.mgh --o $i/surf/rh.Z.thickness.fsaverage.mgh; done

Y for i in `cat list`; do fscalc $i/surf/lh.Y.fsaverage.mgh sub group_mean_X_hc_lh.mgh div group_std_cth_hc_lh.mgh --o $i/surf/lh.Z.fsaverage.mgh; done for i in `cat list; do fscalc $i/surf/rh.Y.fsaverage.mgh sub group_mean_Y_hc_rh.mgh div group_std_Y_hc_rh.mgh --o $i/surf/rh.Z.fsaverage.mgh; done

// Derive paired-difference of Z scores between X and Y

mri_concat --f stacks_XminusY_lh --paired-diff --o concat_XminusY_lh.mgh mri_concat --f stacks_XminusY_rh --paired-diff --o concat_XminusY_rh.mgh

// Smooth concatenated maps of paired differences

mri_surf2surf --s fsaverage --hemi lh --fwhm 4 --sval concat_XminusY_lh.mgh --tval concat_XminusY_lh_sm4.mgh

// Test significance of paired differences within the group

mri_glmfit --y concat_XminusY_lh_sm4.mgh --fsgd fsgd_pairediff dods --osgm --surf fsaverage lh --cortex --glmdir

I used the --osgm flag as there are no covariates here.

---

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