Hi,
Stripes are usually phase wraps and should definitely disappear
after unwrapping. The most likely problem is that the scaling of
the phase is incorrect. The best way to check the scaling is to
find one of the wraps (stripes) and look at the intensity difference
across it. This difference needs to be 2*pi, as that is what all
wraps will be originally. You should be able to figure out
whether the scaling that you've done is appropriate for this.
If you are finding wraps with very different intensity changes
across them (there can be small variations which are OK)
then there must be something unusual with the acquisition itself.
As for more example data - I'm afraid I don't have any to hand.
You might want to also try acquiring real and imaginary data
for comparison, as you can easily convert these to magnitude
and phase using fslcomplex, as:
fslcomplex -complex realvol imagvol complexvol
fslcomplex -realpolar complexvol magvol phasevol
which converts the images realvol and imagvol to magvol and
phasevol. If you then see what this phase is like it may show
you some differences with the previously acquired phase
images.
If you are still stuck and cannot get things to work then feel
free to upload your images to:
http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
All the best,
Mark
On 14 Aug 2009, at 11:45, Virginia Flanagin wrote:
> Hi,
>
> I have been trying to get fieldmaps for fMRI and DTI with our GE
> Signa 3T
> Scanner. I have read all of the threads (particularly from Nov.
> 2007) on the
> difficulties of collecting fieldmaps with the 8-channel head coil
> and have
> tried out most of the different research operations parameters
> (together
> with a GE Technician) on the scanner for dealing with the parallel
> imaging.
> I am able to collect 4 images (magnitude phase real imaginary) for
> each of
> the SPGR sequences and the phase image is consistently between
> -10,000 and
> 10,000, and looks otherwise correct based on the information from
> the fugue
> help. I divided the values in each image by 10,000 and multiplied by
> pi in
> order to get images between -pi and pi.
>
> But the unwrapping with prelude does not seem to work. I still get
> values
> between -pi and pi but the images are still striped in the sagital and
> coronal plane and a lot of them still look like a yin yang in the
> axial
> plane. When I then subjtract the two phase images I get what looks
> like a
> fieldmap in rad/s but there are stripes where the stripes in the two
> images
> did not completely overlap that are either around -1000 or 1000. I
> have
> looked a little bit into the prelude parameters and tried a few of
> them but
> there was never any change. Is there anything that I am doing wrong
> with the
> unwrapping or is this most likely still a GE 3T Asset issue? I
> haven't tried
> the nograd = 1 factor or the rhasset_R but I wanted to make sure
> that the
> problem was still with the images and not something I was doing
> wrong in the
> unwrapping.
>
> I was also just at the FSL course and I downloaded the example data
> sets but
> the phase information is combined in one image from both TEs. Is it
> possible
> to get an example data set where the phase information is separate
> for each TE?
>
> Thanks for your help,
>
> Virginia Flanagin
>
|