[Mrtrix-discussion] b-value correction

[Donald Tournier]

Hi Javier,

Unfortunately there is no simple way to do this within MRtrix at the moment
- this is something that we're working on, and will hopefully be included
in future versions. In the meantime, if you're able to get FSL to rotate
its bvecs file, you might be able to feed that to MRtrix by re-orienting
and re-formatting the resulting bvecs/bvals files into a format suitable
for MRtrix. I don't have anything to do this right now, but maybe someone
else might...?

However, there's a couple of issues I'd like to raise related to your post.
First off, in our experience, eddy_correct does not handle high b-value,
low SNR data very well. What we see is that the DW data are 'stretched out'
(scaled) to match the b=0 image, to a varying degree based on the DW
direction and the corresponding signal around the edge of the brain. We
typically find that it introduces more artefacts than were present in the
data - especially for data acquired using twice-refocused sequences, which
are robust to eddy-current effects (the default on Siemens scanners). That
said, if you're using lower b-values around 1000s/mm², you might find it
works fine - I can only recommend you look closely at the results before
and after correction, we found for example that the FA map was much more
blurred after correction than otherwise.

The other related issue is that as far as I can tell, eddy_correct is not
designed to perform motion correction. Someone may be able to confirm this
for me, since I can't really figure this out from the limited
documentation. If you look at the eddy_correct script, you'll find that it
splits the 4D file into 3D volumes, and calls flirt to realign each volume
to the first one. The issue is that flirt is invoked with the '-nosearch'
option, which according the documentation, "sets all angular search ranges
to 0 0". I can only interpret that to mean that it will do an affine
(scale, translation, shear), but no rotation, and so can't really be used
for motion correction where non-negligible rotations are expected. If
someone has more information about this and can confirm one way or the
other, I'd love to hear about it!

Motion and eddy-current correction on high b-value data is tricky, and
there's been a few abstracts on the topic at recent ISMRM meetings (Jesper
Andersson & David Raffelt at ISMRM 2012, and -I think- someone from Yanniv
Assaf's group at the 2011 ISMRM). Hopefully, we'll have something robust
within the MRtrix pipeline at some point in the not too distant future...

Cheers,

Donald.


On 3 September 2012 19:34, Mr Javier Urriola Yaksic <
javier.urriola at uq.net.au> wrote:

> Dear MRtrix community,
>
> Hi, during the data preprocessing I've corrected head motion and eddy
> current distortions by using eddy_current tool provided in FSL. However, so
> far I haven't  been able to rotate and adjust the b-matrix values in order
> to continue with the MRtrix diffusion processing.
> How can I correct my b-values after eddy_correct registration?
>
> Thanks in advance,
>
> Javier
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>



-- 
*Dr Jacques-Donald Tournier
*
Research Fellow

The Florey Institute of Neuroscience and Mental Health
Melbourne Brain Centre - Austin Campus
245 Burgundy Street
Heidelberg  Vic  3084
Ph:  +61 3 9035 7033
Fax:  +61 3 9035 7307
www.florey.edu.au
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20120904/ab459a1d/attachment.html