Hi,
I've had a quick look at these and I think the registration is in
general pretty good. I suspect the problem of brainstem
mis-registration, which is certainly present, is largely due to
distortions in the diffusion images. Did you apply any form
of distortion correction (e.g. fieldmap-based unwarping)
to these images? I assume they are collected with a standard
EPI sequence. If you can apply some distortion correction
then I would strongly recommend it.
Also, if you can do distortion correction (or it has already been
done) then it might be worth trying to do something about the
edge of the brain mask in the FA (as it contains the
typical noisy, bright voxels). We can introduce a mask around
the edge of the FA to ignore this bright ring and also voxel just
outside this for registration purposes, which makes it less
sensitive to the exact BET mask. However, it isn't really worth
doing this on the distorted data, as the distortions will drown out
this effect.
Finally, if it is only the brainstem you care about, then we can
specify that this is what should be registered well, at the
expense of other areas (which are currently registered quite
well) being made worse.
All the best,
Mark
On 2 Nov 2009, at 21:38, Anastasia A Ford wrote:
> Mark,
>
> Thank you for your reply.
>
> I uploaded T1 and FA images, as well as data, and nodif files. My
> upload
> reference number is 453520.
>
> sincerely,
>
> Anastasia
>
|