Dear Stefano,
There are ways of doing this.
You need to register the fieldmap to the EPI, which will then resample
the fieldmap values to the same resolution as the EPI. This isn't the
only way to do it, but it is the easiest.
To do the registration you should create a forward warped version
of the magnitude-reconstructed fieldmap image and register that to the
EPI. Then apply this transformation to the fieldmap values. This is
described in the lecture of FLIRT/FUGUE in the fslcourse - see the
slides on our website.
Once you've resampled the fieldmap, just use the --loadfmap option
in fugue and things should work OK from there. Note that you'll have
to make sure that your fieldmap is in units of radians per second
(rad/s)
before doing this. You can easily scale it from whatever units the
GE output is by using avwmaths (once you've worked out the scaling
factor).
However, note that this all assumes that your fieldmap is undistorted.
If it suffers from significant distortion (e.g. EPI-based fieldmaps)
then
you need a very different method of dealing with it.
All the best,
Mark
On 22 Mar 2006, at 09:20, Marenco, Stefano (NIH/NIMH) [E] wrote:
> We acquire our own DTI sequence with 2x2 mm resolution with 240 mm
> FOV and 96x96 matrix.
> We then acquire fieldmaps that are reconstructed by the GE software
> with the same FOV but 256x256 or 128x128 matrix.
> The center of the two acquisitions is the same as far as we can
> tell, although there is a half pixel ambiguity.
> Our understanding is that prelude expects the fieldmap and the epi
> sequence to be matched pixel by pixel.
> Is there a way to get around this problem with existing FSL software?
> Thanks a lot, as usual, for your precious advice. Stefano
|