Dear Ged,
thanks for the code and the comments. Apart from my confusion between
tissue density and tissue probability density (as John pointed out), I
agree with you that using the global grey matter image mean, computed
after normalization and using the mean(img(img > mean(img)/8)) formula,
doesn't seem optimal. I see the point, however, in using the full
intracranial volume or the sum of gray and white matter volumes, as a
way of gaining statistical efficiency by partitioning global and local
variance (*both* of which can be of interest).
very best
giuseppe
Ged Ridgway wrote:
> Hi Giuseppe,
>
> I was very interested in your comments on global confounds, I've
> tended to prefer to use total (probable) volumes, which should be
> nearly identical in native and modulated normalised images, instead of
> the spm_global mean(img(img > mean(img)/8)). Just because the latter
> never really made much sense to me, in terms of the arbitrariness of
> the "original mean over 8" concept.
>
> Your example did seem to suggest means might be better; another
> approach would be to use the mean of what actually gets analysed, e.g.
> inside the analysis-mask from an absolute/relative threshold and/or
> explicit mask. I don't think it's clear what the "best" covariate is,
> since the exact clinical question (in terms of "local" effects
> uncorrelated with "global" effects) isn't that precisely defined.
>
> Anyway, back to practical issues...
>
>> 3) if i didn't output the CFS image in the segmentation process (I
>> used the
>> default settings), do I need to run the segmentation all over again to
>> obtain the CSF image and hence the total intracranial volume?
>
> This post:
> http://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0608&L=SPM&P=30287
> with the knowledge that the opts are boolean flags for [mwc wc c]
> output, will hopefully help you here.
>
> Best,
> Ged.
--
-----
Giuseppe Pagnoni
Psychiatry and Behavioral Sciences
Emory University School of Medicine
101 Woodruff Circle, Suite 4000
Atlanta, GA, 30322
tel: 404.712.8431
fax: 404.727.3233
|