Hello,
We were doing some testing between spm5 and spm12 and were unable to run a paired ttest due to different image dimensions post normalization. After some digging around we noticed that when we normalize a given image in spm12 (and spm8 for that matter), using the same input parameters in all cases, an extra slice is removed and added from the image in the y and z dimensions respectively. Since that is hard to describe here are some numbers:
The following were done on images with origin manually set at 31, 30, 15 for all starting points and these were all done in their respective GUIs.
spm5 image origin and dimensions after normalization:
origin: 27,38,18
dim: [-3 0 0 81, 0 3 0 -115, 0 0 3 -53]
Note the whole numbers in the output.
spm8 and spm12 (both old and new norm schemes) image origin and dimensions after normalization:
origin: 27, 38.3, 17.7
dim: [- 3 0 0 81, 0 3 0 -114, 0 0 3 -54]
Note the decimal numbers in the y and z dimensions in the origin and how in the normalization output images the y-dim has one less slice and the z-dim one more. All transform matrices were identical between spm5, spm8 and spm12 (old) normalizations.
Any ideas as to why this may be the case? Thanks,
Keith
|