| In a VBM analysis, it is not quite clear to me whether it is the there
| is a difference in normalising for differences in voxel intensity across
| scans by including the global mean voxel value as confounding covariate
| (Ashburner and Friston, Neuroimage, 2000, 805-21) or dividing the
| partioned images by the relative voxel volumes (May et al., Nature Med,
| 1999, 836-8) (or is this exactly the same ?)
These two approaches both test different things. Modulating the
grey matter images by the deformation Jacobian determinants, results in
the amount of grey matter originating from each brain region being
preserved after spatial normalisation.
Including the global mean as a covariate in the model means that you are
looking for more regional differences in grey matter concentration, beyond
the global differences. This is similar to the old procedures used for
analysing PET data.
|
| Another item I am not quite familiarized with, is whether a zero extent
| threshold can be used for VBM since the statistic based on cluster
| spatial extent is not valid for VBM analysis (again : VBM:The Methods,
| Neuroimage, 2000, 805-12). Nevertheless, in several published articles
| using the technique, the resulting foci were also characterised in
| terms of corrected spatial extent (e.g. Woermann et al, JNNP 2000,
| 162-169; Brain 1999,2101-2107)?
As I understand the extent threshold, it should not enter into the
voxel level statistics.
Best regards,
-John
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|