Hi Avram, it basically looks ok, but I'm confused about the "scanner" regressor. If these are two different MRI scanners, one at 1T and one at 3T, you should probably make different classes for them. Perhaps the same thing for the Console variable.
doug
On 07/15/2013 01:19 PM, Avram Holmes wrote:
Freesurfer et al.,
I'd like to try out some simple cortical thickness regression analysis in mri_glmfit. These would be analyses of one continuous score (e.g., Temperament) and several continuous nuisance variables (e.g., Age, IQ, Site, etc). However, before getting started I wanted to run my fsgd and contrast files by the mailing list to make sure I am entering things correctly. If any freesurfer pros have a free moment I've love to hear if I am on the right track.
My fsgd file looks like the following:
GroupDescriptorFile 1 Title Temp MeasurementName thickness Class Main Variables Temperament Age Scanner Console IQ Input Subj1 Main -2.65 22 3.00 4.00 103.0 Input Subj 2 Main -2.24 24 3.00 3.00 109.0 Input Subj 3 Main -1.82 20 1.00 4.00 121.0 Input Subj 4 Main -1.79 19 3.00 3.00 113.5 Input Subj 5 Main -1.77 19 1.00 3.00 101. 6 ....
My contrast1.mat file is:
0 1 0 0 0 0
It is my hope that when I call:
mri_glmfit --fsgd GLMFIT/fsgd.txt --y GLMFIT/lh.TEMP.mgh --C GLMFIT/c.txt --surf fsaverage lh --cortex --glmdir GLMFIT/lh.TEMP.output
I am prepping to run a regression looking at the relationship between Temperament and cortical thickness, partialing out age, scanner, console, and IQ.
Thanks in advance for any help you can offer, Avram
-- Avram Holmes PhD Instructor in Psychiatry, Harvard Medical School Assistant in Research, Massachusetts General Hospital
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer