George P Thomas Jr wrote:
> Hello community,
>
> Could anyone tell me exactly how you calculate total intracranial volume
Sounds familiar... Not so long ago, some fool had this to say on the
matter... ;-)
www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0608&L=spm&P=742
> Do you simply sum the native segmentation
> volumes (i.e. optimized VBM results)? The white matter in my analyses is
> modulated, so is it proper to sum the modulated WM, GM and CSF volumes to
> arrive at a value?
Modulation should be such that the volume of the modulated warped
segmentation should be exactly the same as the native space
segmentation, since modulation ensures this *locally* everywhere by
multiplying by the Jacobian determinant (which is nothing other than
the volume-change factor of each voxel).
I had a quick glance at this to check that the theory held in
practice, and for a few c1 and mwc1 pairs the differences are about
0.01% of the mean, so nothing to worry about, I'd say.
More of a potential worry with this method is that it's perhaps not
really getting TICV... since the segmentations are probabilistic maps,
and the CSF one may not stop exactly at the dura, and I guess there
maybe other problems. But maybe this is nonetheless a better method my
tape-measure suggestion...
> Furthermore, I am entering TICV as a nuisance variable along with age.
> Would there be a reason to input ICV as a 'user specified global
> calculation' (I'm using the PET, Single Subject, Conditions and Covariates
> model) rather than as I have been doing it?
I'm afraid I have no idea what a 'user specified global
> calculation' is, but I know in the VBM community we use TICV as a
covariate/nuisance variable, so I would think this should be generally
okay...
Best,
Ged.
|