Hi Jason,
The warning is just there to let you know the s-form specifies an
oblique transformation for this dataset. The data is interpreted as
oblique, and this transformation should carry over to the NIFTI
output. The obliquity may explain why the display is slightly
different if FSLview 4.0 interprets this differently than version
3.0. Note you should be able to do this in a single 3dcopy command
with a .nii suffix in the output file name.
Regards,
Daniel Glen
On Apr 10, 2008, at 12:46 PM, Jason Stein wrote:
> Hello FSLers,
>
> I have a question about data formats and nifti headers. I have a
> series of image files, that when opened in fslview using the fsl3.3
> release, there is no problem (meaning that they look like they are
> supposed to and don't give me any errors when opening them).
> However, when I open the same images in fsl4.0 using fslview (or
> any other fsl program) I get the following error
>
> ++ WARNING: nifti_read_buffer(__filename.nii.gz__):
> data bytes needed = 633600
> data bytes input = 633596
> number missing = 4 (set to 0)
>
> And in fslview it appears to be slightly moved over relative to
> where it should be and was in fsl3.3.
>
> I found that a solution to this problem was to convert to AFNI and
> then back to FSL (now the FSL4.0) using the following commands.
>
> #Convert to AFNI
> >3dcopy filename.nii.gz test
> #Convert the AFNI back to NIFTI
> >3dAFNItoNIFTI test+orig. test
>
> When I use the 3dcopy command, I get the following warning message
>
> ++ Program 3dcopy: AFNI version=AFNI_2006_12_22_0933 [32-bit]
> qform not present, sform used.
> sform was not exact, and the worst axis is
> 6.626820 degrees from plumb.
>
> However, my new test.nii now looks just fine when I open it up in
> fslview from the fsl4.0 release and gives me no errors. I believe
> this fixes the problem ok, but I'm not sure what causes the error
> or if it creates some other problems for me with these images. Am
> I creating more problems with this solution? Where does the
> problem come from?
>
> Thanks for any help,
> Jason
>
|