Hi,
We've been experimenting with a few things over the summer
but do not have a single solution that works in all cases. It
depends on how severe the atrophy is, but it can be quite
beneficial to do a non-linear registration to the MNI152
and then apply FIRST to the output. This doesn't always
do a good job and in severe atrophy it isn't great, but it
can really help in a lot of cases.
You will then need to apply the mask to the original image
using an inverse warp. To make sure that this doesn't take
too long, make sure that you crop your original image to
exclude as much non-brain as possible. Then run the
fnirt and the invwarp should be relatively efficient. If you
original image is very high resolution though, it may be
helpful to make a new reference image with the same FOV
but a larger voxel size (1mm or even 2mm) to act in place
of the original image, just for the purposes of invwarp.
Hope this helps.
If you still have problems please let us know, and we'll keep
everyone posted on progress with FIRST and atrophied
brains.
All the best,
Mark
On 9 Sep 2009, at 00:57, Dana Eldreth wrote:
> Dear FSL creators,
>
> I am running FIRST on some older brains that are very
> heterogeneous. For
> most, FIRST does an excellent job, but for some it is substantially
> off because
> of atrophy. Earlier in the summer there was talk of a modification
> to FIRST for
> older brains. I wanted to follow-up to see when that program may be
> available.
>
> I was wondering if there is a protocol recommendation for what
> should be
> done for subjects whose subcortical structures cannot be estimated
> well using
> FIRST. How should this be reported if some subjects need to be
> excluded?
>
> Eagerly waiting FIRST for older brains!
>
> Thanks!
> Dana
>
|