External Email - Use Caution
Have you looked at this page https://secure-web.cisco.com/1HEsOws5HLh6urGfIzkrMbC_7g6t-p8fMP3vYoFDRvAWBsy... ? Bascially I think you will need number of classes = 2=M/F times N=#sights, then have age and age^2 as your continuous regressors. With the FSGD, you have a choice of a full interaction model (DODS=different offset different slope) or have each continuous variable be its own column. What you are describing would be a mixture of these things, so you would need to create your own design in that case. Eg, you could have a column for each class; the value=1 if that subject is in that class or 0 if not. Then you'd have a column for age^2, then you would have two columns for age, one would be the age if the subject is M otherwise 0, the other would be age if the subject is F otherwise 0. You can pass the matrix to mri_glmfit with the --X option (instead of using --fsgd).
On 9/4/2023 3:27 AM, #NGUYEN THUAN TINH# wrote:
External Email - Use Caution
Dear Freesurfer experts,
I want to run a GLM to analyze the correlation between a continuous variable Y and a morphometric measure (eg. Cortical thickness). I want to control for some nuisance variables: sex, scan site, age, age squared, age * sex interaction, head size.
- How should I go about formulating the FSDG file?
- What should my design matrix be if I am only interested in the relationship between Y and the morphometric measure of interest after controlling for the various nuisance variables?
Thank you so much for the help!
Yours,
Sophia
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://secure-web.cisco.com/1ieKsVWqwE1D1cOOd1l0xRXz_ZS9OgIev4aKh4UAVWFQcPf...