Join GitHub today
GitHub is home to over 28 million developers working together to host and review code, manage projects, and build software together.Sign up
Streamline-based Linear Registration #446
This is the first PR for the new crème de la crème method that I presented in ISMRM.
In accordance with @omarocegueda's work you first define a StreamlineDistanceMetric then give this to StreamlineLinearRegistration which has a method called optimize which takes the two bundles (sets of streamlines) as input and returns a StreamlineRegistrationMap which can transform the moving bundle to the static.
I have only one simple tutorial (see bundle_registration.py) for now but I am working with @jchoude to put some streamline datasets online and create a couple more detailed tutorials both for bundle-based, full brain streamline-based registration and bundle-specific atlas creation. But that will come in another PR (probably multiple PRs) most likely after the release.
In all files the coverage should be high. But in optimize.py the coverage cannot be that high because
Also, when you build the modules a new config.pxi file will be created in dipy/build. We use this file to check
@arokem can you prioritize looking at this PR for this release? I will start working on the LiFE PR.
Enjoy your new toy ;)
Thanks for the quick turnaround! I will take a look again in a few days. If
On Thu, Oct 16, 2014 at 11:36 AM, Eleftherios Garyfallidis <
Hi @arokem, I think I answered all your comments and removed also all old matrix44 implementations. Now I use compose_matrix from dipy.core.geometry. I also wrote some tests for compose_matrix and decompose_matrix. Let me know what you think. On my side it seems this baby is ready to be merged.
Looking good. This is real close now - there were just a couple of
On Tue, Oct 28, 2014 at 1:24 PM, Eleftherios Garyfallidis <