[Mrtrix-discussion] Response function rendering

[Donald Tournier]

Hi Javier,

A funny-looking response function is typically a consequence of a
suboptimal 'single-fibre' mask. In your case, your response is actually not
too bad - certainly good enough to get decent results - but yes, it can be
improved.

The issue is that the automated procedure used to derive the 'single-fibre'
mask is admittedly a little rough and ready, so you really need to check
that it's OK before proceeding to the next stage. The main thing to look
out for is that all voxels included in the mask are placed in actual white
matter, in regions that (are thought to) contain single fibre orientations.
It's often necessary to edit this mask to exclude noisy, high FA voxels at
the edge of the brain, etc. You should also modify some of the commands you
use, which will help to produce a more robust mask. I typically use:

$ average img_dwi.nii -axis 3 - | threshold - - | median3D - - | median3D -
img_dwi_mask.mif

to get the whole-brain mask, and:

$ erode img_mask.mif -npass 3 - | erode - - | mrmult img_mask_fa.mif - - |
threshold - -abs 0.7 img_mask_sf.mif

to get the single-fibre mask. Note that these are the commands recommended
in the documentation for the latest release. I will then have a brief look
at the mask overlaid on the FA map just to check there aren't too many
voxels in non-WM regions, and edit if required - it's fine to leave a few
bad voxels in there, no need for it to be perfect. After that, the response
function should be fine.

Hope that helps,

Donald.



On 25 July 2012 15:40, Mr Javier Urriola Yaksic <
javier.urriolayaksic at uq.net.au> wrote:

>  Dear Donald,
>
> I'm new using MRtrix. Therefore, I'm following the online manual as close
> as possible in order to understand the whole process to get reliable
> results.
> While reading the manual I noticed that my CSD response image (attached in
> this mail) looks slightly different compared to the on-line image of the
> manual. The axial plane is larger than the z-axis as expected, yet in the
> centre of the response function rendering is folded slightly outwards (see
> images).
> So far, I haven't figure out what is the cause of this and if it would
> affect my results?
> Thanks!
> Regards,
>
> Javier Urriola Y.
>
>
> Code used:
> $ mrconvert img_dwi.nii -coord 3 0 - | threshold - - | median3D - - |
> median3D - img_dwi_mask.mif
>
> $ dwi2tensor img_dwi.nii -grad  z_iy_xb_dir.txt - | mrmult - img_mask.mif
> img_dtmask.mif
>
> $ tensor2FA img_dtmask.mif - | mrmult - img_mask.mif img_mask_fa.mif
>
> $ tensor2vector img_dtmask.mif - | mrmult - img_mask_fa.mif
> img_mask_ev.mif
>
> erode img_mask.mif - | erode - - | mrmult img_mask_fa.mif - - | threshold
> - -abs 0.7 img_mask_sf.mif
>
> estimate_response  img_dwi.nii img_mask_sf.mif -lmax 6
> img_mask_response.txt -grad  z_iy_xb_dir.txt
>
> disp_profile -response img_mask_response.txt
>
>
>
>
>
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>
>


-- 
Jacques-Donald Tournier (PhD)
Brain Research Institute, Melbourne, Australia
Tel: +61 (0)3 9035 7033
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20120726/d22992a4/attachment.html