Hi Jenessa,
Jenessa Price wrote:
I'm examining differences in cortical thickness in a sample of drug users and non-users. From the literature, it appears that the 2 covariates I should use for the GLM analysis will be gender and age. Other researchers have examined other possible covariates (like age of first use, # past year uses, IQ, etc.) using correlations between those variables and significant clusters that differed between groups. Is this the best way to go about translating a typical regression model into the Freesurfer statistical package (especially if I'm interested in dose-dependent relationships)?
The age, gender, and eTIV (estimated Total Intracranial Volume) are typically used as nuisance variables. If age of first use, etc, are dose-dependent and you're interested in a dose relationship, then I would include one. You can also include more than one and use an F-test.
Also, how do you suggest controlling for total brain volume? Other researchers have performed univariate analyses between groups on TBV controlling for age and gender, then examined mean cortical thickness, but I'm not quite sure what the steps are for this. Could someone help me identify the best way to control for TBV (i.e., is there a way to control from the first GLM analysis, or is this a possible covariate to enter in to the analysis)?
Gender is a good way to control for TBV, since that is the biggest effect. You can also use the eTIV mentioned above to correct for head size. Note that this is a difference between head size and brain size as brain size can shrink with age (and maybe drug use?).
doug
Thanks!
-- Jenessa S. Price, M.A. Graduate Student, Clinical Psychology and Neuropsychology University of Cincinnati Campus Representative, American Psychological Association of Graduate Students (APAGS)
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer