Dear Kristina,
I'm afraid I'm not very familiar with GE scanners and hence have left this in the hope that someone more experienced with GE would reply.
Since that hasn't happened I think I really need to see the scans in order to help.
Can you upload them to:
http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
and send me the reference number?
It sounds like there are still problematic phase wraps in the superior slices, or possibly a problem with FOV.
I'll be able to tell you more when I can see the images.
All the best,
Mark
On 22 Aug 2012, at 11:15, Kristina Skatun <[log in to unmask]> wrote:
> Dear fsl experts,
>
> I tried posting this a few months ago, but still haven't received any response, so I'm trying one more time.
>
> I have a few issues regarding creating field maps and unwarping epi images.
> I use a GE 3T scanner, and the outputs for creating the FMs are real and imaginary images at TE 4.6 and 9.2. Echo spacing: 0.568. EPI TE: 30.
>
> Based on previous posts in this forum on creating FMs from real and imaginary and the fsl webpages, i created the following script:
>
>
> -------------------------------------------------------------------------------------------------------
> #!/bin/csh
>
> # Make fieldmap from real and imaginary images
>
> set current_dir = `pwd`
> set IDs = (O30*)
> set TEs = (TE1 TE2)
>
>
> foreach ID ($IDs)
> cd $current_dir/$ID/Unanalyzed/FM
>
> foreach TE ($TEs)
> cd $current_dir/$ID/Unanalyzed/FM/$TE
>
> # Reorient the images to template orientation
> fslreorient2std notRO_imaginary imaginary
> fslreorient2std notRO_real real
>
> # make complex from real and imaginary images
> fslcomplex -complex real imaginary complex
>
> # Make magnitude (absolute) image of complex img
> fslcomplex -realabs complex magnitude
>
> end
> end
>
> # bet magnitude image and create brain mask (only one mag image is needed)
> foreach ID ($IDs)
> cd $current_dir/$ID/Unanalyzed/FM/TE1
> bet magnitude magnitude_brain -f 0.4 -g 0 -m -R
>
> # copy the magnitude_brain and mask image to the TE2 folder
> cd $current_dir/$ID/Unanalyzed/FM
> cp TE1/magnitude_brain_mask.nii.gz TE2/magnitude_brain_mask.nii.gz
> cp TE1/magnitude_brain.nii.gz TE2/magnitude_brain.nii.gz
>
> # Use prelude (via phase image) to unwrap phase img from complex img using magnitude_brain
> foreach TE ($TEs)
> cd $current_dir/$ID/Unanalyzed/FM/$TE
> fslcomplex -realphase complex phase_rad_$TE
> prelude -a magnitude_brain -p phase_rad_$TE -o uphase_rad_$TE
> end
> end
>
> # Convert unwrapped maps from rad to rad/s by subtracting the two phase maps and dividing by the diff in TE
> foreach ID ($IDs)
> cd $current_dir/$ID/Unanalyzed/FM
> fslmaths TE1/uphase_rad_TE1.nii.gz -sub TE2/uphase_rad_TE2.nii.gz -div 0.0046 fmap_rads_$ID -odt float
>
> echo "--------------------FM for $ID has been created!------------------"
> end
> -----------------------------------------------------------------------------------------------------------
>
>
> All the step until the final one (converting to rad/s) looks ok and the same for all individuals as far as i can tell. One issue i ran into was that i first used prelude to unwrap the complex (and not phase image): prelude -c complex -o uphase_rad -f -v -m magnitude_brain_mask. However, this resulted in that the uphase_rad somehow got flipped so that left and right sides of the brain were opposite that of the magnitude (detected in fslview). Going via phase and then prelude resolved this, though not not sure why.
>
> The final fieldmaps look ok for some of the subjects, however, there are consistent issues that i am not sure how to solve. For close to half of the subjects the superior part of the brain has a marked horizontal divide (in sagital view) of and area of much greater intensity (suddenly changing by more than 1000), like a white top. The size of the area varies from subject to subject, and some don't have it at all. This divide is also apparent in the thresholded signal loss weighting image (red) after unwarping in Feat. Scaling in fslview also varies between subject, but does this have any thing to say?
>
> When unwarping using prestats in Feat, the unwarped images look mainly closer to the magnitude image than the distorted ones. However, the fieldmaps with the divide look more distorted in the superior parts of the brain. In addition, the unwarping shift map in voxels is greater than the recommendation, up to -26 to 28 at the most extreme. I have used the exact same procedure for all subjects and gone over it many times, but cannot find where i am going wrong. Any inputs how to solve this would be most appreciated!
>
> Regards,
> -Kristina
>
|