Re: [Freesurfer] Maps showing correlations between structural measures

I think your command lines are correct. I would not expect that swapping the y and pvr to give the same results. sig=-log10(p)=30 is not totally unreasonable. I don't understand #2. Why are you running FDR and cluster correction?

On 05/11/2017 06:39 PM, Martin Juneja wrote:

In addition to that: I am always getting positive correlations (when I load sig.mgh) every time i.e. even when I correlated thickness and gyrification. None of the voxels/areas showing negative correlation between any of the structural measures

On Thu, May 11, 2017 at 11:26 AM, Martin Juneja <mj70481@gmail.com mailto:mj70481@gmail.com> wrote:

Hi Dr. Greve,

I doubt these results because of following:

(1). When I compare sig.mgh files created using following two
commands i.e. correlation between LGI and CV versus CV and LGI:
mri_glmfit --y Corr_CV_LGI/lh.LGI.10.mgh --fsgd CV_LGI.fsgd dods
--C Corr-VL-cor.mtx --surf fsaverage lh --cortex --glmdir
Corr_CV_LGI/lh.VL.glmdir --pvr Corr_CV_LGI/lh.CV.10.mgh
and
mri_glmfit --y Corr_CV_LGI/lh.vol.10.mgh --fsgd CV_LGI.fsgd dods
--C Corr-VL-cor.mtx --surf fsaverage lh --cortex --glmdir
Corr_CV_LGI/lh.VL.glmdir --pvr Corr_CV_LGI/lh.LGI.10.mgh

(a). I do not get identical maps. Please find attached screen shot.
(b). Here, colorbar showing a value of around 30, does that
reflect -log10(p)? I am not sure if thats normal?

(2). Assuming magnitude from colorbar is normal then I performed
FDR correction- I am assuming its same command as following as
mentioned in FS manual:
mri_glmfit-sim --glmdir Corr_CV_LGI/lh.VL.glmdir --cache 1.3 pos
--cwp 0.01 --2spaces

After running FDR correction, cache.th13.pos.sig.cluster.mgh and
cache.th13.neg.sig.cluster.mgh, none of these are showing me any
significant clusters, instead its showing a constant magnitude
(same color) over whole brain?

Thanks.
​

---

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