[Neuroimaging] Tracking from surface files

[Kristian M. Eschenburg]

Hello

Thanks for the input, both suggestions are promising ideas.  I agree --
projecting the white / gray surface inwards makes sense in order to bypass
the weak diffusion signal at the tissue boundary.  One could then grow the
tracks outwards tangentially to intersect with the actual cortex, or just
define cortical labels on this new surface.


k

On Fri, Aug 26, 2016 at 6:31 PM, Bago <mrbago at gmail.com> wrote:

> Hi all,
>   I believe both of Ariels suggestions are good. Using the mesh vertices
> as indices should be pretty clean, but sampling points on the mesh will
> allow you to more finely control the density and number of seeds. I thing
> worth mentioning is that fiber tracking works best when the seeds are in
> the white matter tissues, so projecting 1 - 2mm from the mesh into the
> white matter (when the mesh is along a white/gray boundary) might make the
> final tacks better.
>
> Bago
>
> On Fri, Aug 26, 2016 at 4:21 PM Ariel Rokem <arokem at gmail.com> wrote:
>
>> Hi Kristian,
>>
>> On Fri, Aug 26, 2016 at 4:51 PM, Kristian M. Eschenburg <keschenb at uw.edu>
>> wrote:
>>
>>> Hello
>>>
>>> I'm interested in tracking streamlines using a mesh like that included
>>> in the HCP files, instead voxels, as seed points. Does Dipy have this
>>> functionality?
>>>
>>
>> Thanks for your email.
>>
>> Seeds in dipy (e.g., inputs to the `LocalTracking` class ` __init__`
>> function; see example:https://github.com/nipy/dipy/blob/master/doc/
>> examples/introduction_to_basic_tracking.py#L98 ) are an array of 3D
>> coordinates, so they don't have to be voxels. But - I don't think we don't
>> have a utility function that does what `seeds_from_mask` does, but for
>> meshes (a seeds_from_mesh function).
>>
>> Do the mesh files from HCP contain the 3D positions of the vertices? You
>> could potentially use the positions of the vertices as seeds. A more
>> sophisticated approach would sample the surface of the mesh between
>> vertices. But I haven't thought about this very deeply, and I'd be curious
>> to hear if others have thought about potential pitfalls with these
>> approaches. If you end up writing something along these lines (even just
>> sampling the vertices) consider contributing that alongside
>> seeds_from_mask!
>>
>> Cheers,
>>
>> Ariel
>>
>>
>>> Thanks!
>>>
>>>
>>> k
>>>
>>> _______________________________________________
>>> Neuroimaging mailing list
>>> Neuroimaging at python.org
>>> https://mail.python.org/mailman/listinfo/neuroimaging
>>>
>>> _______________________________________________
>> Neuroimaging mailing list
>> Neuroimaging at python.org
>> https://mail.python.org/mailman/listinfo/neuroimaging
>>
>
> _______________________________________________
> Neuroimaging mailing list
> Neuroimaging at python.org
> https://mail.python.org/mailman/listinfo/neuroimaging
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20160826/a6505b1f/attachment.html>