Hi Mark,
I've been talking with Tom off-line about this, and here's the solution
that we came up with. I just thought I'd post it in case anyone had any
feedback for us.
thanks
doug
1. Compute the voxel shift map from the B0 map in the highres space
(using prelude/fugue).
2. Align (motion correct) the functional run to the middle time-point
(using mcflirt)
3. Register the B0 magnitude to the middle time-point (flirt)
4. Resample the voxel shift map to the middle time-point based on the
registration. This handles the high-to-low res conversion. (flirt)
5. B0 correct (ie, run fugue) on all time-points of the motion corrected
functional based on the resampled voxel shift map.
Mark Jenkinson wrote:
>
> Hi,
>
> Sorry I haven't replied sooner but I haven't tried this before so I
> needed
> to have a close look at the code. Unfortunately, there is a place in
> the
> code where it assumes the field map and EPI are the same size so you
> can't simply use it as you'd like for now. I will update the code to
> make
> it more flexible and the next release (probably in a few months' time)
> will have this new version.
>
> In the meantime the easiest work around is to down-sample your
> field maps to make them 64x64. I know this isn't ideal but it is easy
> to do and will work pretty well. To do the down-sampling just use
> flirt with something like:
> flirt -ref epi -in fmap -out fmap64 -applyxfm
> This applies a small blurring prior to applying the transformation in
> order
> to maintain information from all voxels in the down-sampling, and will
> output
> a fieldmap (fmap64) which is the same size as the epi, without any
> spatial
> transformation (the default transform for -init is the identity
> transformation).
>
> Then use fugue with the epi and fmap64.
> Note that flirt doesn't work with complex images, so make sure
> the fmap image used above is either the output from fugue with
> the --savefmap option, or the output from prelude.
>
> Also note that it assumes that the images fmap and epi are registered.
> Ideally you should down-sample, forward warp the magnitude image
> from the field map sequence, register this warped image with the epi
> and then apply the transformation to fmap in order to have it properly
> registered with the epi. See the information at
> http://www.fmrib.ox.ac.uk/fslcourse/flirt_fugue
> for more on this.
>
> Hope this does the trick for now.
> All the best,
> Mark
>
> On Thursday, June 3, 2004, at 08:53 pm, Tom Liu wrote:
>
> > Hi,
> >
> > We are starting to look at the use of PRELUDE and FUGUE for
> > unwarping of DICOM EPI images off a GE 3T short bore, and had
> > a quick question: How does FUGUE handle higher resolution field maps?
> > e.g. if we acquire 128x128 map, will FUGUE properly apply it to a
> > 64x64
> > EPI image? Our initial tests seem to indicate that it sort of works,
> > but I haven't
> > seen anything in the documentation or lists indicating that it should.
> > The reason I ask is that the standard SGPR that we're planning
> > on using seems to have a silly lower limit of 128x128, and we'd rather
> > not
> > modify the product code unless we have to.
> >
> > Thanks,
> >
> > Tom Liu
> >
> > ***********************************************************
> > Thomas Liu
> > Center for Functional Magnetic Resonance Imaging
> > University of California, San Diego
> > 9500 Gilman Drive, Mail code 0677
> > La Jolla, CA 92093
> > Phone: (858) 822-0542
> > Fax: (858) 822-0605
> > http://fmriserver.ucsd.edu/ttliu
> > ***********************************************************
--
Douglas N. Greve, Ph.D.
MGH-NMR Center
[log in to unmask]
Phone Number: 617-724-2358
Fax: 617-726-7422
|