Dear Valerie,

It seems that the problem you have stems from the fact that
your fieldmap has a significant non-zero mean.  This *should*
be taken care of inside FEAT but there is a bug which means
that the edges of the fieldmap are corrupted in this case.
We will fix the bug in the next patch.  However, in the meantime
the easiest thing is to ensure that you have a demeaned
fieldmap.  The best way to do this is:
  - run BET on your magnitude image (either one) to make an image
	that I'll call "mag_brain"
 - run the following command to demean (and mask, although that
	is not essential) your fieldmap image:
	fslmaths fieldmap3 -sub `fslstats fieldmap3 -k mag_brain -P 50` -mas mag_brain fmap3_masked

Then use fmap3_masked and mag_brain as inputs in the pre-stats
tab in FEAT.

This should fix your problem.

All the best,
	Mark


On 16 Jan 2011, at 22:14, Valerie Anderson wrote:

> Hi Mark,
> 
> Thanks for your reply.
> 
> I have tried eroding the BETed magnitude image, but still see the same
> jagged edges at the back of the brain.
> 
> I also wondered whether the problem is to do with the brain coverage of
> my fieldmap images, which miss out the top slices of the brain? However,
> if you wouldn't mind having a look at the data I'd really appreciate it.
> I have uploaded the phase image, 2 magnitude images, the fieldmap I
> generated, EPI and structural images - reference number 633912.
> 
> Thanks,
> Valerie
> 
> 
> -----Original Message-----
> From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On
> Behalf Of Mark Jenkinson
> Sent: Friday, 14 January 2011 10:54 p.m.
> To: [log in to unmask]
> Subject: Re: [FSL] B0 unwarping using fieldmap
> 
> Dear Valerie,
> 
> The processing you are doing at the moment all sounds correct.
> We sometimes see problems at the edge of fieldmaps where there is
> noise created by the small partial volume of tissue and skull, and this
> can create jagged edges in the unwarped images.  However, it does
> normally not go away by using the un-BETed image.  This is somewhat
> odd.  I would try eroding your BETed magnitude image by one voxel
> (use fslmaths and -ero) and then see if that fixes the problem.  If not,
> please feel free to upload your data (fieldmaps - phase and magnitude, 
> EPI and structural scan) to:  
>    http://www.fmrib.ox.ac.uk/cgi-bin/upload.cgi
> and send us the reference number, and we'll have a look at it.
> 
> All the best,
> 	Mark
> 
> 
> 
> On 14 Jan 2011, at 03:08, Valerie Anderson wrote:
> 
>> Hi,
>> 
>> I am using Feat pre-stats to try to unwarp EPI images using a
> fieldmap.
>> 
>> I followed the steps on the webpage, but my results aren't right. The
> undistorted example_func image that is output is very jagged/stretched
> at the back of the brain compared with the distorted example_func image.
> Also, the unwarping shift map shows a range of -28 to +10. What I don't
> understand is that when I input a non-BET'ed magnitude image to the GUI,
> the unwarping appeared to work correctly, with the shift map showing a
> sensible range of values (-10 to +10).
>> 
>> Can anyone explain to me what I'm doing wrong please??
>> 
>> The data was acquired on a Siemen's 3T Skyra scanner, and consists of
> 2 magnitude images and what I think is an unwrapped phase difference
> image. I processed the phase difference image to obtain the field map in
> rad/s (using fslmaths), but I didn't use prelude (as I understood that
> the phase difference image was already unwrapped) or do any
> regularisation of the fieldmap. I'm not sure how to determine the 'best'
> regularisation for the acquisition?
>> 
>> Thanks,
>> Valerie
>> 
>