Dear David,
Registering a small Field-Of-View (FOV) to a full brain or large FOV
is a difficult problem and so the standard default settings for
FLIRT won't normally work. To do this kind of registration, where
you have less than a third of the brain, you need to restrict
the degrees of freedom - especially the Anterior-Posterior scaling (the
z-direction for your coronal slices).
To register the partial FOV to a full brain within subject the best
solution is to find the translation fit first then use this to
initialise a small correction in either restricted 2D or 3D.
1) flirt -in partialFOV -ref wholebrain -omat partial2whole.mat
2a) flirt -init partial2whole.mat -nosearch -dof 6 -in partialFOV
-ref wholebrain -omat newpartial2whole.mat -out partial2whole
OR
2b) flirt -init partial2whole.mat -nosearch -2D -in partialFOV -ref
wholebrain -omat newpartial2whole.mat -out partial2whole
To get a good fit to the standard image you need to go via a whole
brain image. That is partial FOV -> whole brain then whole brain ->
standard. The two transformations are then combined to give you
partial FOV -> standard. This process is described in the documentation
on the web and in previous FSL list emails.
If you don't have a whole brain (and I'm not sure if you do or not) then
I doubt you can get very good matches with the standard. We always
recommend that at least one whole brain image is acquired for a subject
and if you are doing partial brain EPI, then an additional whole brain
EPI should be acquired that includes your partial brain slices.
Best of luck,
Mark
David Krauss wrote:
> I want to align 7 subjects to a common space (probably just one of the
> subject's brains). The catch is that our slice prescription only
> includes 12 pseudo-coronal (perpendicular to the calcarine) slices
> extending from the occipital pole to roughly the occipito-parietal
> sulcus. So far, even when I have attempted to register a single brain
> to the built-in template, I have had problems. It appears that FSL does
> not realize the limited slices I have, and the alignment to the
> templates appears improperly warped. So, I need some advice first just
> to properly register each subject to the template, and second to
> register and average all of my subjects in a common space. Any
> suggestions?
>
>
> David Krauss
> Graduate Student
> UCLA Department of Psychology
> [log in to unmask]
>
>
|