[Mrtrix-discussion] extract seed voxels from .tck

[Robert Smith]

Hi there Yu

By default, MRtrix performs *bidirectional* tracking i.e. it tracks both
forwards and backwards from the seed point. To construct a single
continuous streamline from this data, it 'stitches' the two
half-streamlines together - as such, the point along the streamline which
was the seed point cannot be determined. This is partly by design; ideally
the path of a streamline should not be dependent upon exactly which point
along its length was used as the seed, so this information is deemed
irrelevant. It's conceptually possible to modify streamtrack to provide you
with the exact seed point of each streamline in a separate file, but it's
not a trivial change and would require a decent grasp of C++ programming.

By contrast, if you are using *unidirectional* tracking (by specifying the
-unidirectional option), the first point of each track will always be the
seed point. But be careful, the track data provides points in 'real space',
not voxels; these data need to be transformed first if you wish to
determine which voxel each point lies within. The program tracks2prob is an
example of where this sort of transformation is used if you are looking to
implement a similar functionality.

In terms of the random seeding, I won't bore you with the statistics but as
long as your number of streamlines is much greater than your number of seed
voxels, you can expect at least one streamline to be seeded from each
voxel. But bear in mind that the random seeding is not only a selection of
which voxel to seed from; it also randomly selects a position within the
volume occupied by that voxel. This is important as a streamline seeded
from one corner of a voxel could traverse a very different path to one
seeded from the opposing corner. You could consider the ratio between your
number of streamlines, and the number of voxels in your seed, as a sort of
'over-sampling ratio', defining the approximate number of streamlines which
are seeded from different starting points within each voxel. The greater
this ratio (i.e. the more tracks you generate), the less likely you are to
'miss' potential pathways due to the random nature of the sampling.

Hope this is of some help to you
Rob

--

Robert Smith
Melbourne Brain Centre
245 Burgundy Street
Heidelberg VIC 3084
Telephone: (+61 3) 9035 7128
Fax: (+61 3) 9035 7301
Email:  r.smith at brain.org.au
www.florey.edu.au <http://www.fni.edu.au>        www.brain.org.au



On Fri, Dec 9, 2011 at 9:25 PM, yu zhang <zhangyu2ustc at gmail.com> wrote:

> Dear all:
>
>   I wanted to use the streamtrack to generate the connectivity pattern of
> the seed mask.
>   I have extracted the streamlines from the .tck files. It included 5000
> streamlines without any information about seed voxels.
> At first, I assumed that the first voxel in track.data was the seed voxel.
> However, the generated seed mask was far different from the original seed
> mask which I had used.
>
> I was wondering how could I define the seed voxel for each streamline?
>
> Besides, how could I ensure that the random seeds starting from each voxel
> in the seed mask?
>
> Thank you very much!
>
> Sincerely,
>   Yu Zhang.
>
>
> _______________________________________________
> Mrtrix-discussion mailing list
> Mrtrix-discussion at www.nitrc.org
> http://www.nitrc.org/mailman/listinfo/mrtrix-discussion
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.nitrc.org/pipermail/mrtrix-discussion/attachments/20111210/51197240/attachment.html