Hi,
It is a little hard to see the result clearly from the image.
It does look like the edges do not align well over portions
of the brain where you do not expect to see significant
distortions in the EPI (e.g. superior regions). If this is
the case then something has not worked well. The main
registrations to check are the example_func to highres
and the highres to standard. All other registrations are
formed from these, so if either of these is bad then the
others will be affected.
The first thing to do is to determine which registration
is failing. Then this will help identify the cause. For
example, if either the EPI or the T1 image has been
reconstructed with the wrong voxel sizes, then the 6 DOF
registration would fail, but this would need to be fixed at
the reconstruction stage. However, this is only one of
many possible causes, so please check these registrations
and then get back to us.
All the best,
Mark
On 16 Mar 2009, at 20:36, shih-Wei Wu wrote:
> Dear all,
>
> I have a question about FLIRT. I used 6DOF to the subject's high res
> T1
> (after applying bet with -B -f 0.4 -g 0) and then 12DOF linear to
> standard
> MNI space (default option).
>
> When I look at the results (attached here), from example_func to
> standard,
> it looks like
> there are some mismatches between the red contours and the image,
> especially
> in the second row. In both the sagittal and coronal slices, the
> upper edge
> of the red contour fell below the upper edge of the standard image.
>
> Is this considered good/successful registration? If not, how can I
> improve it?
>
> Thanks!
> shihwei
> <example_func2standard.jpg>
|