Hi Mike/Katherine,
I'm a little unsure as to what you mean by "slices ... fall out".
Do you mean that the image cuts off before the cerebellum
(ie. the field of view of the image is smaller than ideal) or
do you mean that the bottom slices of your image, where the
cerebellum should be, are dark and have virtually no intensity
due to bias field or other effects?
In the former case it should not affect the registration. FLIRT
does not assume that the entire brain is present in both images
and can work with one or both of the images being partial fields
of view (although the smaller the FOV the worse the registration
gets, but a few cms of the inferior brain shouldn't matter much).
In the latter case it is a serious confound for FLIRT as it will
view the dark bottom slices as some boundary which is only present
in one image and therefore cannot match well with the other image.
In this case it is better to remove these bottom slices, or
use a cost function weighting image that is zero for these slices
and one for all slices with good intensity.
If neither of these seems to be the case then please email us
back and explain what is going wrong in more detail.
All the best,
Mark
On 20 Oct 2004, at 23:17, Katherine Turner wrote:
> A colleague of mine is having trouble getting on the listserv and asked
> me to post this question for him. He can be reached at
> [log in to unmask]
>
> ********
> Due to problems with field of view, the inferior slices of some
> anatomical images in my data fall out (some portions of cerebellum for
> example, cannot be detected well). This has been causing rotations in
> my
> images after using FLIRT for registration and resulting in
> misalignments
> between my stuctural and functional images.
>
> What is the best way to treat this data in FLIRT so that I can account
> for these missing structural data and keep my functionals from being
> rotated to misalignment during linear registration?
>
>
> Thanks,
> Mike Rubio
|