Could it be possible that someone "fixed" the origin in the various images
released with SPM? I would suggest using one of the overlay functions
available via the results part of SPM. See if those regions that appear
outside the "glass brain" seem to be in a similar location in the tpm data.
If there is a mis-match between the tpm data and the glass-brain, then this
may indicate some sort of corruption of the tpm.
Alternatively, take a look at the contrast image to see how well it
corresponds with the t statistic image. Because the t stat is proportional
to the contrast image divided by the square root of the residuals, it is
possible that the maxima in the t statistic become shifted towards regions of
lower residual variance (ie further away from the brain). In VBM studies,
the maxima of the t statistic field are not usually located at the same
points as the maxima in the contrast field.
Best regards,
-John
On Friday 11 April 2008 10:44, Carl Johan Ekman wrote:
> Philipp,
> Thank you for helping me! Yes, I have compared the smwc*.img files with
> the templates in the tmp folder. When segmenting though, I use 1x1x1 mm3
> voxel size setting, with the distributed templates, can this cause a
> problem?
> I followed yor suggestions and checked the unsmoothed images, to me they
> seem to be aligned. I attach the CheckReg picture containing (from left to
> right, top to bottom):
> 1. The Grey template from the tpm folder.
> 2-8. 8 unsmoothed images selected at random.
> 9-13. The Beta images.
> 14. The Con image.
> 15. The Mask image.
> Can you find anything weird?
>
> Best regards,
> Carl Johan
|