Check whether rh.ad_hv.thick.10.mgh and rh.ad_hv.area.10.mgh are the same, ie,
mri_diff  rh.ad_hv.thick.10.mgh  rh.ad_hv.area.10.mgh

If those are the same, then check whether thickness.fwhm10.fsaverage and area.fwhm10.fsaverage  are the same in each subject

On 3/14/16 10:53 AM, Kaiming Yin wrote:
Dear Douglas,

My command lines were, e.g. right thickness and right white surface (using the same ad_hv.fsgd and group_diff.mtx files in the command lines):


ad_hv.fsgd

GroupDescriptorFile 1
Title AD_HV_GroupAnalysis
Class AD
Class HV
Input AD_003 AD
Input AD_005 AD
...
Input HV_AB HV
Input HV_AE HV


group_diff.mtx

1 -1


right thickness

mris_preproc --fsgd ad_hv.fsgd --cache-in thickness.fwhm10.fsaverage --target fsaverage --hemi rh --out rh.ad_hv.thick.10.mgh

mri_glmfit --y rh.ad_hv.thick.10.mgh --fsgd ad_hv.fsgd dods --C group_diff.mtx --surf fsaverage rh --cortex --glmdir rh.ad_hv.thick.glmdir

mri_glmfit-sim --glmdir rh.ad_hv.thick.glmdir --cache 4 neg --cwp 0.05 --2spaces


right white surface

mris_preproc --fsgd ad_hv.fsgd --cache-in area.fwhm10.fsaverage --target fsaverage --hemi rh --out rh.ad_hv.area.10.mgh

mri_glmfit --y rh.ad_hv.area.10.mgh --fsgd ad_hv.fsgd dods --C group_diff.mtx --surf fsaverage rh --cortex --glmdir rh.ad_hv.area.glmdir

mri_glmfit-sim --glmdir rh.ad_hv.area.glmdir --cache 4 neg --cwp 0.05 --2spaces


Their results were all the same.

Thanks,
Kaiming


---------- Forwarded message ----------
From: Douglas N Greve <greve@nmr.mgh.harvard.edu>
To: freesurfer@nmr.mgh.harvard.edu
Cc: 
Date: Mon, 7 Mar 2016 18:47:11 -0500
Subject: Re: [Freesurfer] group analysis all measures same results
what are your command lines?


On 5 March 2016 at 02:01, Kaiming Yin <yinkaiming6987@gmail.com> wrote:
Dear guys,

I was doing group analysis (command-line) on two groups (patients and control, no age or other information). The data has been "qcached", and after following the three steps as "mris_preproc, mri_glmfit and mri_glmfit-sim" shown in the tutorial on the Freesurfer website, I obtained the difference on thickness in both left and right hemispheres, which seems great. However, when I repeated the above steps on other measures, e.g. area, area.pial, volume (also replaced the parts of "thickness" in the command lines), the results seemed to be exactly the same again as those in the thickness comparison, even the p-value numbers in the file "cache.th40.neg.sig.cluster.summary" were the same. I was using Freesurfer v6.0-beta, and I performed the individual data analysis on a linux cluster and I copied/downloaded the whole "qcached" subject directory to my linux laptop to perform the group comparison, both using the same version of Freesurfer, does it matter? Or do you know why it happened like this?

Thanks,
Kaiming