We have done a thorough examination through trial and error to attempt to
solve this problem.
We changed the headers on the spiral scans so that the were in
register with the anatomical scan. We then went through the realignment,
which worked well. The images were still within register with the
anatomic. We then coregistered the two images:
Target=T1 (28 slices)
Object=first spiral scan (20 slices)
The spiral scan was moved to a position that we believe is the same as the
result after coregistration without editing headers. The problem is that
the scans are not in register following coregistration in the z direction.
We also tried coregistration with:
Target=spiral scan
Object=T1
We also determined normalization parameters from the T1 and applied them
to the poorly coregistered spiral image to see if normalization would
're-register' the spiral image to the 'correct' orientation.
We also tried the mutual information option for coregistration with the
same poor result.
All the poor results are basically the same
Any further suggestion would be greatly appreciated,
Jason and Dane.
On Wed, 29 Mar 2000, John Ashburner wrote:
> The <Check Reg> button is probably the best way of ensuring that
> the images are in register. You can select a number of different
> images, which will be displayed together as SPM understands their
> relative orientations to be. You can click on different parts of the
> orthogonal views, which will re-centre the cross-hairs.
>
> The <Coregister> button should have a good go at registering the images
> together. However, if this does not work well (e.g., because of low
> grey/white contrast in your images), then you can try the Mutual Information
> coregistration that you can get at via the <Defaults> button.
>
> Before coregistering the images, the origin should be set to be somewhere
> close to the AC (which can be done via the <Display> button). Repositioning
> images in this way changes their .mat files. By default, SPM reads voxel
> sizes and origins from the .mat files. However, if an image has no .mat
> file, then the values are read from the image .hdr. The origin field
> of the .hdr should contain the co-ordinate (rounded to the nearest voxel,
> where the first voxel is at 1,1,1) of a voxel reasonably close to the AC.
> After spatial normalisation, the origin is at the voxel that the spatial
> normalisation deemed to be closest to the AC. There is more documentation
> on this stuff available via the <Help> button (Data Formats).
>
> Best regards,
> -John
>
> | Here is what I have:
> | A T1 anatomical scan of 28 slices
> |
> | Spiral functional scans of 20 slices
> |
> | How can I ensure that the spiral scans are coregistered in the proper
> | slice planes with the T1 anatomical scans?
> |
> |
> | Second question:
> | How is the origin field determined?
>
Jason R. Steffener, RTSV
Department of Psychiatry, University Hospital, UMDNJ
Office: (973) 972-1604
http://www.umdnj.edu/~steffejr/jason.html
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|