ENH: Add n4_only workflow -- to skip brain extraction

[adelavega]

Allows skipping brain extraction with n4_only workflow.

See: nipreps/smriprep#116

I think the main remaining todo is this comment from @oesteban:

I've been looking at some masks after binarization. I think they would benefit a lot from a binary opening followed by closing to remove the irregular boundary. Also, it would also make sense to fill holes. We might want to implement that as part of the Binarize interface. I'll have a look into this tomorrow.

[pull-assistant]

Best reviewed: commit by commit

---

Optimal code review plan

     Add init_n4_only_wf

     Fix docstring

     Apply suggestions from code review

     FIX: Connection input to binarize

     Update niworkflows/anat/ants.py

     out_mask for Binarize thr_brainmask

Powered by Pull Assistant. Last update 2ce0b5b ... 298f067. Read the comment docs.

[codecov]

Codecov Report

Merging #435 into master will decrease coverage by 4.3%.
The diff coverage is 9.52%.

@@            Coverage Diff             @@
##           master     #435      +/-   ##
==========================================
- Coverage   59.98%   55.68%   -4.31%     
==========================================
  Files          57       57              
  Lines        5448     5675     +227     
  Branches      644      647       +3     
==========================================
- Hits         3268     3160     -108     
- Misses       2137     2456     +319     
- Partials       43       59      +16

Flag	Coverage Δ
#masks	?
#reportlettests	100% <ø> (ø)	⬆️
#unittests	43.52% <9.52%> (-0.1%)	⬇️

Impacted Files	Coverage Δ
niworkflows/anat/ants.py	12.15% <9.52%> (-0.35%)	⬇️
niworkflows/func/tests/test_util.py	0% <0%> (-100%)	⬇️
niworkflows/func/util.py	28% <0%> (-52%)	⬇️
niworkflows/interfaces/fixes.py	47.05% <0%> (-39.22%)	⬇️
niworkflows/engine/workflows.py	26.31% <0%> (-15.79%)	⬇️
niworkflows/interfaces/registration.py	52.15% <0%> (-9.92%)	⬇️
niworkflows/viz/utils.py	74.09% <0%> (-7.52%)	⬇️
niworkflows/interfaces/utils.py	40.71% <0%> (-7.39%)	⬇️
niworkflows/interfaces/ants.py	67.7% <0%> (-6.25%)	⬇️
niworkflows/interfaces/images.py	47.47% <0%> (-4.78%)	⬇️
... and 3 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6ff8a54...298f067. Read the comment docs.

[oesteban]

Just to avoid future iterations to add the binary dilation, let's do it now.

[oesteban]

Just one detail left...

[adelavega]

Should the dilated mask be output, or the undilated one (which is what we're doing here)

[oesteban]

I'd say the undilated. I believe we should refactor the Binarize interface so that it can do the binary closing and opening, which will give a smoother mask (and that will be good for the dilated mask too).