Hi Alexandre
It looks like your phase-offset maps (the _rad_s.jpg files) are
corrupted - they should look smooth within the brain. Those
discontinuities you have in the phase-offset maps will easily explain
the corrupted final image.
So then we need to work out why your phase-offset maps go wrong...
The thing that seems most suspect to me is the very first stage in
your task list where you appear to divide the measured phase by the
maximum amplitude recorded. Do you know what units the phase image is
in when you get it from the scanner? Our Siemens systems give an image
with units 0->4096 corresponding to -pi to pi phase offset which we
then need to unwrap.
The phase input to prelude should look mostly smooth, but have a few
discontinuities which prelude can unwrap. If you look at the pre-
prelude phase image in fslview you should be able to see that the
phase jumps by exactly 2*pi at these discontinuities. If it changes by
a different amount, then you need to verify each step of the scaling
you've made up until that point. If the phase jumps are 2*pi then
prelude should work...
Hope this helps...
Dan
On 27 Nov 2008, at 14:36, SUBSCRIBE FSL Anonymous wrote:
> Dear all,
>
> We tried to use FSL for fMRI phase map corrections. However, we
> encounter
> some troubles as you can see on the images we posted here:
> http://dl.free.fr/uJwchrtHB/fMRI_Phase_Map.zip.
> Here is the list of the tasks we performed:
>
>> fslstats subject_TE10_phase.nii.gz -R -> get the |max| A of the
>> images
>> fslmaths -dt float subject_TE10_phase.nii.gz -div A.0 -mul 3.1415927
> subject_TE10_phase_div -odt float
>> prelude -a subject_TE10_magnitude.nii.gz -p
>> subject_TE10_phase_div.nii.gz
> -o subject_TE10_phase_unwrapped_rad
>
> Same with TE15
>
>> fslmaths subject_TE15_phase_unwrapped_rad.nii.gz -sub
> subject_TE10_phase_unwrapped_rad.nii.gz -div 0.005 phase_rad_s
>> fugue -i epi.nii.gz --dwell=0.000896 --loadfmap=phase_rad_s.nii.gz -u
> epi_corrected
>
> We would be much grateful if you could explain us what is the
> artefacts on
> the snapshots and how to correct them.
>
> Thanks in advance for your help,
> Best regards,
>
> Alexandre.
>
|