> I have structural MR images of two AD groups (treated and placebo) acquired
> at baseline, and follow up. I processed my longitudinal data following
> Chetelat's protocol (Neuroimage 2005) with SPM2 preserving "total", so I
> will analyze GM volume loss.
I don't actually know what this protocol involves, but I'm guessing that the
procedure is intended to produce "modulated" spatially normalised grey
matter.
> In prospective studies, I usually include TIV
> as nuisance in order to correct for head size. How can I include this in
> the longitudinal model "Compare group: condition & covariates"?
You would use some sort of paired t-test kind of design for this. I don't
know which stats options you would use to obtain this though.
> May I
> choose "the proportional scaling" ("Global normalization Ancova" reduces
> too much the degrees of freedom). Analyses carried out choosing "prop.
> scaling" or "no global normalization" are completely different, which do
> you think is the most correct to use? Thanks, bye
The models ask different questions of the data. Proportional scaling works
with volumes as a proportion of intra-crainal volume, whereas the ancova
model localises differences that could not be explained by intra-cranial
volume differences. I started a WIKI for VBM. It still needs a lot of work,
but it may have some useful information in it.
http://en.wikibooks.org/wiki/SPM-VBM#Globals
Best regards,
-John
|