External Email - Use Caution
Hi Doug, Thank you for the advice! I tried plugging the output of aparc2stats into mri_glmfit both in tandem with —y (mri_preproc_output).mgh as well as on its own and got an error message from mri_reshape saying the elements cannot be changed nv1=163842, nv1=2. This was an iteration of the table with one data column for the thickness of an ROI, however, I also got the same error with the full table. This is also the same table we used in SPSS for a linear regression there.
Terminal inputs: mri_glmfit —y lh.thickness.inflammatory.10.mgh —fsgd inflammatory_markers.fsgd —C Inflammatory_group.mtx —C group_inflammatory.mtx —surf fsaverage —table entorhinal_table.dat —label entorhinal.label —glmdir Inflammatory_entorhinal.lh.thickness.10.glmdir
mri_glmfit —fsgd inflammatory_markers.fsgd —C Inflammatory_group.mtx —C group_inflammatory.mtx —surf fsaverage —table aparc_table.dat —label entorhinal.label —glmdir Inflammatory_entorhinal.lh.thickness.10.glmdir
Thanks! Erin
On Feb 18, 2021, at 11:15 AM, Erin Moe Erin.Moe@colorado.edu wrote:
Hi Freesurfer experts, I’ve been running mri_glmfit and QDEC to investigate the effect of inflammatory markers on cortical thickness and don’t really get much. However, when I extract the cortical thickness values (with aparc/asegstats2table) and run a linear regression with those same inflammatory markers in SPSS, I get several significant results. I’m assuming it might be due to me setting up the QDEC/mri_glmfit analysis wrong.
For QDEC I have a two column .dat file with the subject IDs in the FSID column and the inflammatory markers for each participant in the other column and running the QDEC thickness analysis on that and I don’t get much in terms of results.
For mri_glmfit I used an FSGD file set up just like the .dat file for QDEC with the exception of formatting. However, we wanted to constrain the analysis to an ROI (entorhinal in this case), I ran mris_preproc with the following command: mris_preproc —fsgd inflammatory_markers.fsgd —cache-in thickness.fwhm10.fsaverage —target fsaverage —hemi lh —out lh.thickness.inflammatory.10.mgh
Then fed that output into mri_glmfit as: mri_glmfit —y lh.thickness.inflammatory.10.mgh —fsgd inflammatory_markers.fsgd —C Inflammatory_group.mtx —C group_inflammatory.mtx —surf fsaverage —label entorhinal.label —glmdir Inflammatory_entorhinal.lh.thickness.10.glmdir The results from this look the same as the QDEC analysis but just masked out for the entorhinal ROI. So, to constrain the analysis to the ROI, I ran a Monte Carlo correction with mri_surfcluster.
So a few questions: Would running mri_surfcluster on the sig.mgh output from the mri_glmfit analysis actually constrain the analysis to the ROI specified?
How can I get my results to match that of what I’m seeing in SPSS?
Is my analysis set up correctly for what I’m trying to do?
Thank you in advance for your patience with this lengthy inquiry.
Best, Erin