Attached is a version of eddy_correct that incorporates Martin's bvecs
correction code as an optional argument. This way, the bvecs correction
occurs in one step at the same time as the eddy_current correction. The
code works as normal if you don't specify a fourth argument (the bvecs
file).
As far as the difference it makes: the mean FA difference is 0.000316 for
the dataset I am currently working with. From visual inspection of the V1
lines, it doesn't make any significant difference there either. Perhaps, as
you say, it only matters for datasets with ALOT of subject head motion.
Peace,
Matt.
-----Original Message-----
From: FSL - FMRIB's Software Library [mailto:[log in to unmask]] On Behalf
Of Steve Smith
Sent: Saturday, March 28, 2009 5:57 AM
To: [log in to unmask]
Subject: Re: [FSL] Donating scripts for rotating bvecs after ecc
Thanks very much Martin! This is great - can we get feedback from
anyone who has tested this and seen how much difference it makes?
Do people think that it is likely that datasets with large enough
subject motion to benefit from this correction would also be the same
datasets that are still problematic because of _within_ image
distortions from head motion? Or are the "thresholds" for defining
these cases at different levels?
Cheers, Steve.
On 25 Mar 2009, at 19:40, Martin Kavec wrote:
> Hello to all FSL-ers,
>
> I would like to provide for public use two bash scripts (attached),
> which I
> wrote for the purpose of rotating the bvecs after ecc/motion
> correction in
> FDT. Most of the people (including me) neglect this issue. However,
> it was
> brought to the attention by people around Prof. Derek Jones on the
> ISMRM 2008
> that particularly in tracking application a substantial error may
> accumulate
> along distance of the track. I was also motivated to write the
> scripts, when
> this fact was pointed out by a reviewer of my manuscript.
>
> The first script (ecclog2mat.sh) takes as an input the log file
> resulting from
> ecc/motion correction and generates number of the FSL mat files and
> a list of
> these files with default name mat.list.
>
> The second script (rotbvecs) is based on the script written by Matt
> Glasser,
> (thanks Matt), and it takes as arguments the original bvecs, the
> name of a
> new_bvecs and the mat.list file, which is an output from ecclog2mat.
>
> I have tested the functionality of the scripts by rotating motion
> corrected
> data, nodif, and nodif_brain_mask using Nudge by relatively large
> angles,
> about 25 degrees in x, y, z, xy, xz, yz, and xyz directions, and fed
> the
> rotbvecs with the multiple matrices (# = number of diffusion
> direction),
> obtained from Nudge.
>
> Saad, took the trouble looking at the scripts and results of the
> Nudge (thanks
> again Saad), and he agrees that the scripts should be doing the job.
> We had an
> idea that there could a radio button (possibly default On) in the
> ecc/motion
> correction GUI calling this functionality automatically.
>
> It would be interesting to compare bedpostx results with and without
> rotation
> of the bvecs, but I haven't had the time to look at that yet.
>
> Looking forward to hearing your comments and happy tracking.
>
> Martin
> <rotbvecs.tar.bz2>
---------------------------------------------------------------------------
Stephen M. Smith, Professor of Biomedical Engineering
Associate Director, Oxford University FMRIB Centre
FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
---------------------------------------------------------------------------
|