[Mrtrix-discussion] Tracking parameters for high resolution monkey data

[Donald Tournier]

Hi Cibu,

Many of these considerations would be discussed in our recent MRtrix paper:

*Tournier J, Calamante F, Connelly A. *
*MRtrix: Diffusion tractography in crossing fiber regions. *
*International Journal of Imaging Systems and Technology 2012;22:53–66. *


I would personally discourage the use of SD_STREAM, for a number of reasons
outlined in that paper. I would always recommend the use of SD_PROB since
it is designed to deal with uncertainty, which is a prominent feature of
tractography.

Otherwise, I'd recommend a step size of approximately 1/10th the voxel
size, with a radius of curvature of around 5x the step size for SD_PROB (it
will need to be much smaller than that for SD_STREAM). A cutoff of 0.1
typically works well, but will depend on the quality of your data - if you
have very high SNR data, you may be able to lower that value somewhat. The
minlength depends to some extent on your research question, but as a
general rule I think a minlength of a few voxels is a good idea to reduce
the incidence of short artefactual tracks arising from noisy regions, where
a spurious FOD peak might make it above threshold over a short distance.

Finally, the number of tracks is always dependent on the research question.
In general, more tracks are always preferred, especially with the
probabilistic approaches, with the only limitation being the obvious
increased computation and storage requirements. The only recommendation I
can make here is to keep increasing until the results appear sufficiently
stable for your application. To generate a TDI map using a whole-brain
approach, I say the 3 million that you're using sounds like it should be
about right. For a simple visualisation of whole-brain tracking, you might
want to reduce this somewhat to ensure the tracks can actually be loaded
into the viewer. For more targeted tracking, you should be able to use much
lower numbers, depending on the exact pathways of interest. In short, as
you can see this is very much application-dependent.

Hope this helps,

Donald.


On 20 September 2012 06:23, Thomas, Cibu (NIH/NICHD) [V] <
thomascp at mail.nih.gov> wrote:

> Hello
> I am trying tractography on ex-vivo macaque brain high angular DW data
> acquired at a resolution of 250 microns isotropic.
> I am using the following command to generate tracts in the whole brain
> streamtrack SD_STREAM CSD8.mif -seed mask.mif -mask mask.mif cst_csd.tck
> -step 0.0375 -curvature 0.049 -grad grads.txt -number 3000000 -minlength 2
> -cutoff 0.10
>
> My question is, are there any recommendations regarding the optimal
> parameters to be used for tracking with the high res data that I have.
> Specifically, given the voxels size of the data, is the step size,
> curvature, number, minlength and cutoff appropriate? Is there an
> anatomically/empirically guided way to assign these parameters? I couldn't
> find any in the literature. What is a reasonable rule of thumb?
>
> Thank you
>
> Cibu Thomas
> _______________________________________________
> 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/20120920/516315b9/attachment.html