ENH: Add ITK-LTA conversion test #66
Conversation
|
Best reviewed: commit by commit
Optimal code review plan
|
|
Hello @oesteban! Thanks for updating this PR. There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻 Comment last updated at 2020-03-19 15:22:14 UTC |
oesteban
force-pushed
the
test/lta-itk-conversion
branch
from
970da02
to
a1ce70c
Compare
Codecov Report
@@ Coverage Diff @@
## master #66 +/- ##
=======================================
Coverage 98.63% 98.63%
=======================================
Files 11 11
Lines 950 951 +1
Branches 129 129
=======================================
+ Hits 937 938 +1
Misses 7 7
Partials 6 6
Continue to review full report at Codecov.
|
Also, finally fixes nipy#64.
In other words, the ``src volume`` is the moving image and the ``dest volume`` is the reference image. However, the matrix written in the file seems to map coordinates from ``src`` to ``dest`` (i.e., it is the right one to move the data from ``dest`` to ``src``.
oesteban
force-pushed
the
test/lta-itk-conversion
branch
from
00f47cb
to
772a791
Compare
| @@ -1,30 +1,28 @@ | |||
| # transform file /home/oesteban/tmp/fmriprep-ds005/fprep-work/fmriprep_wf/single_subject_01_wf/anat_preproc_wf/surface_recon_wf/fsnative2t1w_xfm/T1_robustreg.lta | |||
| # created by oesteban on Sat Mar 14 19:28:37 2020 | |||
|
|
|||
There was a problem hiding this comment.
Does this new file fail prior to #65?
There was a problem hiding this comment.
Yes, I have updated both because I had messed up when filing #64. The attached ITK file (in that issue) I believe is the inverse.
To be sure, I searched again for both (LTA, ITK) from the robust register node and subsequent nodes and replaced them.
This PR starts the implementation of "collapsing" series of transforms, for the linear case. It is tested on real data added in the context of nipy#66, nipy#74 and nipy#75. For the moment does not cover unintended use-cases such as collapsing transform sequences containing one or more nonlinear transforms. Resolves: nipy#88
Summary
This PR adds a few tests, and also addresses an uncovered cause of #64.
This PR assumes that the matrix found in LTA transforms is given in
mode-pointsinstead ofmode-images(in BIDS-Derivatives terms).Further investigation of this would be advised, this is just the empirical result of using fMRIPrep-generated LTAs (and ITK conversions) for the new tests.
TODO
I have added an FSL affine generated from an LTA (bbregister) - Since we don't have references for it, I couldn't add a test this time.