[Mrtrix-discussion] questions about streamtrack command

Donald Tournier d.tournier at brain.org.au
Wed May 23 17:40:41 PDT 2012


Hi Marina,

How the command streamtrack choose the seed point?
>

Seed points are generated at random by uniform sampling from the mask. The
mask is defined as any location in 3D space where the image intensity after
interpolation is greater than 0.5. You can disable the mask interpolation
step with the -nomaskinterp option, so that seeds are only placed strictly
within the voxels of interest. Unfortunately there is no option currently
to ensure regular seed placement.


> Morever, while the command is running, I notice that there are generated
> streamlines but only some of them are selected, and the number of selected
> streamlines corresponds to the value given at the -number parameter.
> Which is the criterium by which streamlines are selected?
>

Streamlines are selected if they satisfy all constraints. By default, there
is a minimum length constraint, which I guess is the only reason you would
currently get different numbers of generated vs. selected tracks. You can
change that using the -minlength option. In general, additional constraints
can be placed using the -include & -exclude options.


> Is there an option to set the number of generated streamlines instead of
> the number of selected ones?
> I've tried to run the command without any options but the numbers of
> generated tracks and selected ones do not match.
>

There is no explicit option, but same result can be achieved very easily by
setting -maxnum and -number to the same value (by default, maxnum is
100xnumber).


>
> My aim is to have a certain number N of tracks beginning from each seed
> point (voxel with intensity equal to 1) of the mask.
>

This is a common request, but it's still not possible to do very easily. If
the number of streamlines generated is large, you can certainly set the
number of tracks to num_voxels x num_tracks_per_voxels, and get a very good
approximation to uniform sampling of your ROI (you can find out the number
of voxels in the ROI using the mrstats command). If you need to save the
tracking results for each voxel separately, then you'll need to break up
your ROI into individual voxel ROIs, and launch one streamtrack command per
voxel. This can easily be done with a bit of scripting, but there is no
easy way within MRtrix to break up the ROI into individual voxel ROIs.
Maybe this is something that I could provide, since it would be trivial to
code up. I'll have a think about that...

Hope that helps,

Donald.



> Thank you for your helpfulness,
> Marina_______________________________________________
> 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/20120524/50a50518/attachment.html


More information about the Mrtrix-discussion mailing list