Hi Nicole,
First of all, doing a single slice to volume registration is *really*
difficult. We strongly recommend that if you are acquiring a
single-slice functional run then you also acquire one whole brain
volume with the *same* sequence as the functional (normally EPI)
and make sure that it is oriented in the way so that one of its slices
is the same as your functional slices. It is better if it has the same
resolution and same distortions as your functional images. In that
way it is much easier to match the single functional slice to this
whole-brain EPI, and then the whole-brain EPI to the structural, etc.
This is what is set up to run straightforwardly in FEAT with the
"initial structural image". Note that it isn't really a "structural" in
this case, as this is more of a historical name.
If you do not have this then you will need to run FLIRT yourself
in order to try and get a good registration of your single-slice
functional to the structural image. This will almost certainly require
some time in order to get it right. If the orientations are roughly the
same, then the 3 dof option provided by the GUI (or with the
xyztrans.sch schedule file) might get you close, and then you can
try to refine this using it as a initial estimate (-init) and doing a
6 dof registration with limited search (-nosearch). However, this
can be quite tricky from a distorted sequence like EPI to a non-
distorted
anatomical one. I'm not sure how your high resolution spiral
acquisition
fits into this. If it looks more like your functional images, then
this is
the best target to try.
Another thing is that all first level FEAT analyses are done in the
native
functional space. So you should not apply the registration to the 4D
time series. What FEAT needs is the registration matrices in order to
resample all the resulting stats into standard space for higher level
analyses. Once you've got a good registration then you just replace
the appropriate matrices in the xxx.feat/reg directory and run
updatefeatreg, which will create all the concatenated matrices,
inverse matrices and images for the feat report pages. The critical
matrices that you need to replace are:
example_func2initial_highres.mat
initial_highres2highres.mat
highres2standard.mat
assuming that you are using the spiral as the "initial highres".
Otherwise
it will be:
example_func2highres.mat
highres2standard.mat
I hope this helps.
All the best,
Mark
On 20 Nov 2008, at 01:33, Nicole Pelot wrote:
> I've acquired single-slice data, and would like to run a FEAT FMRI
> Analysis
> on it. However, the registration through FEAT does not seem to work.
> As a
> trouble-shooting step, I successfully registered a certain five-
> slice data
> set, and then reduced that set to a single-slice; it would no longer
> register. I've been using an initial high resolution full-brain spiral
> acquisition, followed by a full-brain anatomical image, and finally,
> the
> standard brain.
>
> I looked into using FLIRT 2D registration. However, I've had two
> issues.
> First of all, I don't understand how using FLIRT separately would
> tie into
> my FEAT analysis. Second of all, when I did give FLIRT a try, it
> reduced my
> functional data sets down to one volume.
>
> Any help would be very much appreciated! Thank you,
>
> Nicole
>
|