My best guess is that you did something wrong somewhere. The headers
of the original images should not change when you use them to estimate
the normalising transformation via the Normalise button. Changes to
the header may occur when you coregister the data, although this
depends on what you consider to be the moved and fixed images. If
image 1 is coregistered to match image 2, then the header of image 1
will be modified to reflect its new position. However, if image 2 is
coregistered to match image 1, then the header of image 2 is modified.
Best regards,
-John
On 10 November 2011 11:15, Elisa Filevich <[log in to unmask]> wrote:
> Dear experts,
> I was comparing the results of an analysis done in SPM5 with my analysis of
> the same data done in SPM8.
> The results differed (heavily) after preprocessing so I tried to track down
> the differences.
> I noticed that immediately after doing the DICOM import, the structural
> images in SPM8 and SPM5 were exactly the same. However, the same structural
> images, after coregistration and normalization the SPM8 images were
> different from those in SPM5.
> Even more oddly, the date and time of the .img file corresponds to that of
> the DICOM import. But the time of the header file corresponds to that of the
> normalization. If I replace the header file with the original header,
> obtained just after importing, then again the SPM8 and SPM5 images match.
> Is this reasonable?
> If the header files are replaced, does it mean that I can't go back to a
> previous step and repeat the analysis?
> Could I solve this if I used NIfTI format instead?
> Many thanks,
> Elisa
|