search

MRtrix3 / mrtrix3

MRtrix3 provides a set of tools to perform various advanced diffusion MRI analyses, including constrained spherical deconvolution (CSD), probabilistic tractography, track-density imaging, and apparent fibre density
http://www.mrtrix.org
Mozilla Public License 2.0
291 stars 179 forks source link

Change name of FACT tracking algorithm #39

Closed Lestropie closed 10 years ago

Lestropie commented 10 years ago

Technically MRtrix's tensor tracking isn't actually FACT, it's Euler integration. FACT implies nearest-neighbour interpolation.

jdtournier commented 10 years ago

Well, FACT has become the umbrella term for deterministic tensor streamlines. I appreciate that Sumumu Mori's original implementation used nearest-neighbour interpolation, but I'm pretty sure every other version since probably uses some other form of interpolation. And I don't think the specific type of interpolation makes any difference to the overall concept behind FACT, so I think it is the right term - and it's nice and short, which is much easier on the command-line...

If you really weren't happy about that term, the most appropriate would be 'deterministic tensor streamlines' - a bit too long as you might appreciate. Euler integration is a generic numerical integration method, which is not specific to dMRI, so I think that term would just confuse users. Personally, I'm happy with FACT...

Lestropie commented 10 years ago

Well, sort of. But it's irked me for a while, and FACT being the umbrella term is a little erroneous if you don't appreciate the details.

For instance someone might use it and cite Susumu's paper, when in reality our implementation is closer to the Basser / Conturo papers. Also deterministic streamlines with multi-fibre models are almost always advertised as an extension of FACT; in part implying the lack of interpolation. But yeah, I'm stuck on a substitute; although det_tensor and prob_tensor would probably suffice.

jdtournier commented 10 years ago

OK, let's roll with det_tensor and prob_tensor, they are definitely more explicit and less prone to misinterpretation, while still being relatively short for command-line. Happy enough with that - it's not like I'm likely to be typing that too often anyway... ;)