Hi Nissen,
> In VBM analysis, ANCOVA is often used to remove global difference by
> inclusion of the global mean voxel value as a confounding
> covariate. Does it mean to include mean intensitiy of each scan as nusiance
> variable in a ANCOVA statistic procedure? If yes, does it also mean to
> derive (calculate) mean intensities of scans respectively for use?
More meaningful "globals" for VBM are obtained from the the actual tissue
volumes, rather than the globals computed by SPM. If you want to use an
ANCOVA model, then I would suggest not using the global correction built into
SPM, but instead include a list of volumes as a confounding covariate.
All the best,
-John
|