[Neuroimaging] [dipy] Understanding the ACT tracking

[Samuel St-Jean]

Hello,

I ran the example about the new tracking API over here [1] but there is a
small note at the bottom which I don't really understand. The note reads as


Notes

Currently in ACT the proposed method that cuts streamlines going through
subcortical gray matter regions is not implemented. The backtracking
technique for streamlines reaching INVALIDPOINT is not implemented either.

So for the second part, it does no do backtracking, but what about the
first sentence? Should one read it as if does not keep streamlines going
form GM to GM? So I tried the example on the given dataset, and I get
strange results. No idea if pictures work in mailing list, so I'll link to
the issue [2].

I was expecting the ACT tracking to not produce small, anatomically
implausible streamlines, but I'm wondering if it has something to do with
the note above. Anyone can share experience about the results they have? I
tried on an HCP dataset and I also get weird results. I also ran the same
tracking parameters/same masks on another tracking script implementing ACT
(from which a subpart of it is what became the tracking API in dipy, or at
least the same guy did help in both), and it does not produce spurious
tracks as I would expect.

So, am I misunderstanding the ACT tracking and it is supposed to do that or
I am missing something obvious?


[1]
http://nipy.org/dipy/examples_built/tracking_tissue_classifier.html#example-tracking-tissue-classifier

[2] https://github.com/nipy/dipy/issues/831
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/neuroimaging/attachments/20160118/a6b137d9/attachment.html>