Hi Susie,
If you are not getting any image output then it probably is failing,
as this is really a very tough problem. If the reverse registrations
seem to be working OK then that's fine - there is no reason that
these can't work in reverse, although we normally find that the
correlation ratio prefers the "better" image to be the reference,
but this is not a hard and fast rule.
What I would recommend is using 6 DOF though, as 12 could
very easily produce extremely implausible stretches of the slice
to match the 3D volume. So as long as you are happy with the
results (please inspect them carefully) and have used 6 DOF
then I think that's all fine.
In future though I would definitely recommend acquiring a single
volume of spin-echo EPI with the same orientation (including this
same slice amongst others) in order to get better registration.
Should only take a few seconds more scan time!
All the best,
Mark
On 11 Nov 2008, at 03:28, Susie Heo wrote:
> Mark,
> The single slice is from a spin-echo EPI readout and the fMRI whole-
> head 4D
> is from a gradient-echo EPI. Unfortunately, all the data have been
> collected already so I don't have the option of taking a whole brain
> volume
> for each subject using the spin-echo EPI sequence. I've tried playing
> around with different registrations using FLIRT which have yielded the
> following results:
> 1. spin-echo EPI to fMRI to MPRAGE using 6 DOF = no image
> 2. spin-echo EPI to fMRI to MPRAGE using 12 DOF = no image
> 3. spin-echo EPI to MPRAGE using 6/12 DOF = no images
> 4. MPRAGE to fMRI to spin-echo EPI using 6 DOF = single slice high
> res image
> 5. MPRAGE to spin-echo EPI using 6 DOF = single slice high res image
>
> It seems like #4 is more accurate than #5 when overlaid with the
> original
> slice but it is difficult to tell.
>
> Is it defensible to register a high res image to a low res slice??
> What is
> actually going on here? And why isn't there an image when I try to
> use the
> suggested registration sequence (slice to fMRI to high res / slice
> to high res)?
>
> Susie
>
|