Timm and Steve,
Thanks for your help with the fieldmap distortion correction.
In trying out your suggestions weve made changes to so the fieldmap and
fieldmap mag images/files are in the same alignment & orientation, the
qform/sform entries are identical, and tried usesqform in the FLIRT
command line. We also discovered that we were using an input file in error,
and corrected that too.
We found after registering the magnitude and fieldmap images to example func
the fieldmap was not well registered, so we used applied the FLIRT transform
for the magnitude image to the fieldmap.
After correcting these errors, our FEAT output still does not look right.
Im attaching a portion of the pre-stats report file. I'm also attaching
images of the filedmap and magnitude files that we inputed to FEAT and
exfunc image we used for registration in flirt.
Again, we would appreciate any input you might have.
tx
Greg
Quoting Steve Smith <[log in to unmask]>:
> HI - yes, it looks also possibly like the fieldmap and fieldmap mag
> images are not in the same space with the same alignment and
> orientation, and possibly are even entered into the GUI the wrong way
> round?
>
> Cheers.
>
>
> On 26 Nov 2007, at 08:42, Timm Wetzel wrote:
>
> > On 25.11.2007, at 00:14, gtau wrote:
> >
> >> Im writing to get some help with registration.
> > [...]
> >> I then deskulled and registered it to example func running the FEAT
> >> GUI.
> >> FEAT completed running but in the report, the registration is
> >> really off.
> >
> > Your FEAT analysis is overlaying sagittal and coronal images (best
> > seen in the second and second to last rows).
> >
> > I've seen a very similar effect with FLIRT, and it seems that in my
> > case it was caused by unexpected orientation info in the NiFTI header.
> >
> > Does the FLIRT part of your processing complain about the input
> > images having both qform and sform entries in the NiFTI header? (You
> > might need to run it with `-v´.)
> > If yes, I'd either try to use another NiFTI converter or add `-
> > usesqform´ to the FLIRT command line.
> >
> > You can manually check the orientation info with fslhd; in my case
> > FreeSurfer's mri_convert apparently generally fills both the qform
> > and sform entries with the same data.
> >
> >
> > Regards,
> > Timm
|