Hi all,
Sorry for the follow up but hope someone can clarify following doubts.
Thanks, Sneha ________________________________ From: freesurfer-bounces@nmr.mgh.harvard.edu [freesurfer-bounces@nmr.mgh.harvard.edu] on behalf of Sneha Pandya [snp2003@med.cornell.edu] Sent: Tuesday, October 07, 2014 1:06 PM To: freesurfer@nmr.mgh.harvard.edu Subject: [Freesurfer] Study design for qdec or command line group analysis
Dear freesurfers,
I have a study in which we want to perform group analysis to find the effect of age on cortical thickness between patients and controls. We have 11 patients with age group from 45-75 years and the mean of 53. We shall age match our controls with respect to similar age range and mean.
However our patient scans are from both 1.5T and 3.0T scanner, hence I have few questions about designing our study.
A) Can we treat certain discrete factors as continuous variables, like gender (male with value 1 and female with value 2), and scanner (1.5T with value 1 and 3.0T with value 2) to make group analysis qdec-able in case of 3 discrete factors? I tried to look into mailing archive to figure out the answer, but so far found confounding answers.
B) If what asked above is possible can you please suggest which design I should opt for? Design 1) Three discrete factors, each with 2 levels: (i) gender (male and female) (ii) groups (healthy and patients) (iii) scanner (1.5T and 3.0T), Continuous variable: (i) age
OR
Design 2) Two discrete factors, each with 2 levels (i) gender (male and female) (ii) groups (healthy and patients) Continuous variables: (i) age (ii) scanner (1.5T as 1, and 3.0T as 0)
From literature we are not expecting any effect of gender and scanner on change of cortical thickness with age between groups, however I guess we need to control for it right?
Should I implement design 1 with mri_glmfit command line, as it has 3 discrete factors or design 2 in qdec?
B) We are expecting some change in cortical thickness with age between groups. Hence we shall be using DODS model to perform the analysis. In this case I should demean my ages right (subtract grand mean of age from combined healthy and patient groups and subtract from every individual age)?
Thanks, Sneha