Various things could maybe help here:
1) Make sure you have the latest updates to SPM5.
2) Ensure that the starting estimates are not too far out. Sometimes, you can
reorient one of your images to avoid the same local optimum.
3) Maybe try tweeking the various settings to see if you can find one that
works well for your data.
4) Sometimes, a bias correction helps. If you can segment the images, then
there is an option to produce bias corrected versions.
5) Note that if your EPI are heavily distorted then it is not possible to
obtain an accurate alignment using only a rigid body transform. The FieldMap
toolbox could help here.
6) If your images are extremely low quality, then you may need to resort to
some form of manual alignment, or some alternative registration procedure.
Best regards,
-John
On Tuesday 16 September 2008 17:26, Huang, Wei (Psychiatry) wrote:
> Hi, All,
>
>
>
> I posted this last time but didn't get a resolving answer, and now I'm
> getting the similar problem again and hope somebody can help:
>
>
>
> I did the realignment of my functional EPI series of 200 scans with the
> dimension of 128x128x24, and now I coregistered the 240x240x210 T1 3D
> anatomical to the mean of EPIs. The registration result doesn't look
> too good. According to my eyes there's at least some translation in
> both X and Y directions needed. Anybody knows why or how to make it
> better?
>
>
>
> Thanks,
>
>
>
> Wei Huang
|