I would probably correct total GM, total WM, and total ventricular CSF by head size. It is not surprising that they covary a lot (as you point out). But what you want is what is left over after you remove the effect. You can also divide by the eTIV rather than regressing it out, but the results will be similar.
As for thickness, many people regress out the mean thickness, but this changes the hypothesis that you are testing (and so the interpretation of the results). As such, this is a question that you will have to answer yourself.
On 4/9/18 10:11 AM, Maria Gloria Rossetti wrote:
Dear freesurfers
my study aims to measure volumetric differences between patients X and controls. As dependent variables, I have (i) a-priori ROIs and (ii) global brain measures (total gm, total wm and CFS). In the ROIs analysis, I use eTVI to control for for head size variability. Now my question is: should I control for head size when measuring overall brain atrophy? Theoretically speaking I'm not sure. Practically speaking there is a high positive correlation (as expected) between global brain measures and eTVI in my sample so I think it wouldn't be correct to use eTVI. But still, is global atrophy affected by head size? should I control for it?
Second quick question: I'm planning to replicate the analysis measuring thickness differences. Should I control for tot mean thickness then?
Thanks in advance for your help.
Gloria
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer