I would guess that this is the usual problem in which the registration
finds a local optimum. Before running Coreg, try doing a Check Reg to
see if the initial alignment is within about 3 to 4 cm, and about 10
degrees. If it is not, you may need to reposition some of the images
manually, using Display. This involves entering appropriate
translations (in mm) and rotations (in radians), clicking the
"Reorient images..." button, and selecting the images that need to be
moved.
As a reminder, the registration routines in SPM are all susceptible to
local optima. Registration involves searching for parameter values
that best align the image. Each time a set of new parameters is
tried, one of the images needs to be resliced in order to compute the
objective function. Local optimisation methods are used to search for
the best solution. These are a bit like finding the top of a mountain
wearing a blindfold, simply by climbing uphill until you can't get any
higher. Start in the wrong place, and you don't end up at the top.
The alternative would be to try a range of parameter values and
selecting those that give the best solution. If 100 different values
for each parameter are used, for 6-parameter rigid-body registration
this would require one of the image to be resliced a trillion times.
This would take a while, so instead it is done using a local
optimisation.
Normally, if the DICOM conversion routines are used (and scans were
collected within the same scanning session without repositioning the
subject), the initial alignment is close enough for this to work.
Best regards,
-John
On 27 April 2011 16:10, Carolina Valencia <[log in to unmask]> wrote:
> Hi SPM users,
>
> I have a problem with the coregistration step failed and a point of the functional image does not match to a point in the structural image. (image attached)
> I did all the preprocessing steps including the reorientation of the structural and functional images according to the AC-PC axe. I realized during this process in the matlab main window appears the following message, I check the structural image and the origin correspond to the AC point.
>
> Thanks,
>
> Best Regards,
>
> Carolina
>
> Running 'Display Image'
> Done 'Display Image'
> Done
>
> ??? Attempt to reference field of non-structure array.
>
> Error in ==> spm_image at 213
> tmp = spm_get_space([st.vols{1}.fname ',' num2str(st.vols{1}.n)]);
>
> ??? Error while evaluating uicontrol Callback
>
|