The problem arises when you have two images that SPM thinks are in very
different positions. This can be because the .mat files or .hdr files
contain information telling SPM that this is the case. Try using the
"Check Reg" button to see what is happening. If the images appear very
far apart, then delete any .mat files that may be associated with the
images. If "Check Reg" still shows problems, then reset the origin fields
of the images so that the field is the same in all cases.
The coregistration works by minimising the sum of squares difference
between the images. This uses a first estimate, which is then
iteratively repeated until the sum of squares can not be decreased. If
the images begin too far apart, then this procedure fails.
All the best,
-John
| I tried to coregister a structural MR(T1) to a mean PET image.
| MR:256*256*24,set origin [0 0 0]
| PET:128*128*35,set origin [64 64 17] or [0 0 0]
| But I failed in this process whether I set origin of PET image to [64 64 17] or
| [0 0 0].Then I got the following error massage.
|
| To my strange, I successed this coregistration using SPM96.
| Please tell me how to deal with this problem.
|
| Thank you for any help.
|
|
| SPM99b: spm_coregister (v2.5) 18:16:43 - 28/10/1999
| ========================================================================
| Coregistering Subject 1: First Pass Coregistration -
| fine..
| * - SPM99b: spm_affsub3 -------------------------------------------
|
| There is not enough overlap in the
| images to obtain a solution.
|
| Please check that your header information is OK.
|
| ----------------------------------------- 18:20:52 - 28/10/1999
|
| ??? Error using ==> spm_affsub3 (spm_affsub1)
| There is not enough overlap of the images to obtain a solution
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|