Hi,
If all the images are within-subject then I would use 6 (or maybe 7) DOF
and no more. Using 12 DOF is likely to distort the images
unrealistically.
However, getting a good match between a single slice and a full image
is hard - very hard if the single slice is an EPI. We generally
recommend
that people get a single whole brain volume with their EPI (even if they
are only interested in one slice) as one single volume is quick to
acquire and much, much easier to register. It is also helpful to get
fieldmaps in order to undistort the EPI. Also, is your EPI a spin-echo
one (typical for diffusion) or gradient-echo (typical for FMRI)? The
signal loss in gradient-echo EPI makes registration harder with
fieldmaps. If it is the same as your FMRI then I would suggest
registering it to that first and then registering that to the MPRAGE,
assuming that the FMRI is not single slice (as you said 4D). This
should help, although a very different angled slice will still be
quite difficult.
I hope this helps.
All the best,
Mark
On 10 Nov 2008, at 17:47, Susie Heo wrote:
> Hi,
> I would like to register one slice of a subject's low res EPI data
> to his high res MPRAGE,
> however I'm not sure what the best strategy would be. The major
> issue is that the single
> slice EPI was taken at a different angle than the MPRAGE. The
> images that I have are:
> 1) Single slice EPI at level of hippocampus taken at an unknown
> angle (oriented so that it
> spans the length of the hippocampus)
> 2) MPRAGE with slices parallel to AC-PC axis
> 3) fMRI 4D functional image with slices parallel to AC-PC axis
>
> It was suggested that I register the MPRAGE to the single slice EPI
> using FLIRT w/ 12 DOF,
> but I do not know if I should register a high res image to a low res
> one, and with affine
> transformation at that.
>
> Any help (with corresponding rationale) would be greatly appreciated.
>
> Susie
>
|