Hello,
YES, if you did n o t normalise the images in your first type of analysis then
this is of of course a BIG difference, not to say, you cannot perform voxel based statistics
on these images then. Your point no 1 reads like ii you produced segmentated images
in the native space. For a voxelwise comparison corresponding voxels need to come to lie
in the same position.
The Gaer-Scripts (optimised VBM) of course include a normalisation step.
So, either after your segmentation step in native space you normalise
the grey matter to the grey matter template of SPM2
or
you continue to use the optimised VBM scripts you mention,
or you use SPM5 . Here, you could just use the segment funciton but define that you want normalised and modulated images
These mwc...-images need can then be smoothed (e. g. 12 mm) and be used for an ad hoc VBM comparison.
The results from this procedure should at least roughly match those of the optimised VBM scripts.
Best regards,
Philipp
|