Hi Philipp
Again than you for your comments. I indeed suspect from looking more
thoroughly at the data that in the technically proper analysis some of
the GM effect has been shifted to WM.
Accordingly I suspect that it may be helpful to do e.g., a multivariate
test on GM + WM. This means looking at individual voxels which show both
a GM and WM effect. I guess this would be ok because due to the
smoothing each voxel can represent some GM and WM. However, it might be
more optimal to have some across voxel test in which voxel clusters are
computed with voxels having a GM and/OR WM effect.
Kind regards, Dennis
-----Oorspronkelijk bericht-----
Van: Philipp Saemann [mailto:[log in to unmask]]
Verzonden: dinsdag 26 juli 2005 17:24
Aan: Ent D.vant
CC: [log in to unmask]
Onderwerp: Re: [SPM] Christian Gaser VBM scripts with "preserve total"
Hello Dennis,
I guess you have noticed John's answer saying that modulation while
producing
the priors is not recommended.
I do not have a really logical explanation for the discrepancy you
describe
between the results using different priors.
However, the difference must lie in the segmentation step.
You could investigate if your GM was (mis)classified as WM in the
technically
proper analysis - either by looking at WM maps or summing modulated WM
and GM
maps. We observed this type of 'segmentation shift' in Huntington"s
disease
when deep grey matter nuclei correctly 'disappeared' but re-appeared as
increased WM volume.
Best regards,
Philipp G. Saemann
Max-Planck-Institute of Psychiatry
NMR Study Group
Kraepelinstr 2-10
80804 Munich
|