Dear Liat,
it you perform fieldmap correction and don't "see much change" there
are two things that can potentially have gone wrong.
1. Your calculation of the field my be scaled wrongly.
2. You enter the wrong dwell time, i.e. the scaling Hz->distortions is
wrong.
For the first option check that your fieldmaps have "reasonable"
values. For a 1.5T scanner you expect to see an off-resonance field in
the order of ~50Hz around the sinuses area. This scales with the field
so you should expect twice that for a 3T scanner. If you see values
much smaller than this it indicates you have a problem with your
scaling, which leads to "under correction" of the distortions.
A "reasonable" dwell time is on the order of ~0.5ms. If you have a
value much smaller than this it will also lead to "under correction".
Finally, it is sometimes that case that visually it looks visually
like an area suffer from geometric distortions (e.g. the edge of the
frontal lobe has been "pushed back into the brain") and you assume you
should be able to correct it using a fieldmap. Where as in reality the
same area also suffers from signal dropout which means there is little/
no signal to reassign.
Good luck Jesper
On 19 Jul 2010, at 10:48, Liat wrote:
> Dear FSL,
> To try and reduce signal drop out in OFC, we have scanning on the
> coronal plan with an Epi sequence, this indeed results in better
> signal from ventral parts of the brain, but results in severe
> distortions in temporal regions We have been trying to correct
> distortion in our Epi images using field maps. However, when we run
> the distortion correction via Fugue, we don't actually see much
> change I am not sure what we maybe doing wrong when we try and
> correct for the distortions we see, and it is very possible that the
> parameters for our field maps acquisitions may not be the optimal
> ones. I am not sure if there is a way to upload actual data files,
> for you to examine. I tried to enclose the results (as a jpg
> picture) but that also exceeded upload limit set. Hence, I just
> pasted the script we used to derive the phase and magnitude
> infomation from our fields maps, which we acquired with two
> different TE lengths 6600 msec and 7200 msec.
>
> Your help and advice will be much appreciated,
>
> Many thanks.
>
> Regards,
>
> Liat
> ------------------------------
> 1. Script for converting our fieldmaps
>
> #! /usr/bin/python
>
> import os, sys
>
> real1 = sys.argv[1]
> imag1 = sys.argv[2]
> real2 = sys.argv[3]
> imag2 = sys.argv[4]
> TE = str(sys.argv[5])
> fieldmap_out = sys.argv[6]
>
> VERBOSE=True
>
> def runcmd(cmd):
> if VERBOSE:
> print cmd
>
> return os.system(cmd)
>
> #Step 0, generate complex images from real and imaginary data
> runcmd('fslcomplex -complex ' + real1 + ' ' + imag1 + '
> temp_complex_acq1')
> runcmd('fslcomplex -complex ' + real2 + ' ' + imag2 + '
> temp_complex_acq2')
>
> #Step 1, generate magnitude images from complex
> runcmd('fslcomplex -realabs temp_complex_acq1.nii.gz
> temp_fieldmap_mag1')
> runcmd('fslcomplex -realabs temp_complex_acq2.nii.gz
> temp_fieldmap_mag2')
>
> #Step 2, generate phase images in radians
> runcmd('fslcomplex -realphase temp_complex_acq1.nii.gz
> temp_phase0_rad_orig 0 1')
> runcmd('fslcomplex -realphase temp_complex_acq2.nii.gz
> temp_phase1_rad_orig 1 1')
>
> #Step 2b, rescale the phase images to a range 0 - 6.28 (not -3.14 -
> +3.14)
> runcmd('fslmaths temp_phase0_rad_orig.nii.gz -add 3.14159265
> temp_phase0_rad.nii.gz')
> runcmd('fslmaths temp_phase1_rad_orig.nii.gz -add 3.14159265
> temp_phase1_rad.nii.gz')
>
> #Step 3, create phase unwrapped images
> runcmd('prelude -v -a temp_fieldmap_mag1.nii.gz -p
> temp_phase0_rad.nii.gz -o temp_phase0_unwrapped_rad')
> runcmd('prelude -v -a temp_fieldmap_mag2.nii.gz -p
> temp_phase1_rad.nii.gz -o temp_phase1_unwrapped_rad')
>
> #Step 4, generate field maps in rad/s
> runcmd('fslmaths temp_phase1_unwrapped_rad.nii.gz -sub
> temp_phase0_unwrapped_rad.nii.gz -mul 1000 -div '+TE+'
> temp_fieldmap_rads -odt float')
>
> #Step 5, apply fieldmap regularisation
> runcmd('fugue --loadfmap=temp_fieldmap_rads -s 1 --
> savefmap='+fieldmap_out)
> runcmd('fugue --loadfmap='+fieldmap_out+' --despike --
> savefmap='+fieldmap_out)
> runcmd('fugue --loadfmap='+fieldmap_out+' -m --
> savefmap='+fieldmap_out)
>
> #fugue -i episplit0000.nii.gz --dwell=0.394 --loadfmap=out.nii.gz -u
> unwarped.nii.gz --saveshift=shift.nii.gz --despike
>
|