Feature request: read VTK format streamlines

[neurolabusc]

Hello-
The current version imports MRTrix TCK and TrackVis TRK format files. It would be great if one could import VTK format files, as this seems to be the most popular format across tools (DTI-TK, Fiber Navigator, Camino, MedInria).

[jchoude]

Hi,

yes, that would indeed be a good addition to the tool. We are not sure of the time needed to add such loader (because of all other projects needing our time).

In the meantime, if you already have some python-based workflows, you could use Dipy, which now has a function to compress streamlines that uses the exact same algorithm has this binary tool. You can see an example here, in the Dipy documentation for streamlines compression

To be able to load your VTK streamlines to process them with Dipy, you can use the TractConverter

If you need help with those, please let us know!

[neurolabusc]

Thanks for your quick response. Perhaps you could add an example script to
http://nipy.org/dipy/reference/dipy.tracking.html#id40
that shows how to load streamlines from a popular format, compress them and then save them. I assume that you could save the simplified tracks to a popular format (e.g. VTK), but would see even smaller files by using the arithmetic or Huffman encoding of the fib format. I think a simple example would help - DiPy has a terrific collection of tools, but for a new user it is daunting to see how to use the functions together.

[StongeEtienne]

Hi neurolabusc,
Sorry for this late answer.
Be carefull if your converting your .vtk to another format because most of your header will be lost.
I think Dipy support a ".fib" format which is actually a renamed ".vtk"

If not, I did a Pull request recently to add "vtkPolyData loader in Dipy"
dipy/dipy#1034
there is also a function : lines = get_polydata_lines(line_polydata)
and after you should be able to use Dipy compression:
http://nipy.org/dipy/reference/dipy.tracking.html#dipy.tracking.benchmarks.bench_streamline.bench_compress_streamlines

[jchoude]

@StongeEtienne Not sure exactly why you're saying that.

We normally encourage people to reuse the same file format for the input and output. However, when writing a vtk file, the minimal header information needed to work is kept.

If you feel other fields need to be kept, you are more than welcome to make a PR :)