ENH: Verify derivatives are compatible with anatomical reference

[mgxd]

Closes #458

This was partially addressed with masks in #443. This implementation generalizes across files (mask, aseg) and across surface reconstruction methods.

[codecov]

Codecov Report

❌ Patch coverage is 75.34247% with 18 lines in your changes missing coverage. Please review.
✅ Project coverage is 56.74%. Comparing base (1e6afd4) to head (d216490).
⚠️ Report is 48 commits behind head on master.

Files with missing lines	Patch %	Lines
nibabies/workflows/anatomical/fit.py	5.26%	18 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #459      +/-   ##
==========================================
+ Coverage   56.61%   56.74%   +0.12%     
==========================================
  Files          70       70              
  Lines        6747     6788      +41     
  Branches      828      831       +3     
==========================================
+ Hits         3820     3852      +32     
- Misses       2718     2727       +9     
  Partials      209      209              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[mgxd]

@xxnagit would you mind testing this patch to verify it fixes your issue?

https://github.com/nipreps/nibabies/wiki/Testing

[xxnagit]

I have to run this pipeline on HPC, if there is any docker image available, I'm more than happy to try it out. Thank you so much!

[mgxd]

sure, you can use nipreps/nibabies:derivatives-no-preproc

[xxnagit]

I tried a few more subjects this time as the first time when I reported the bug it was just for one subject. I realized my input are quite different for the orientations, and the outputs of Bibsnet are not uniform due to the consistency with the original input. While when these derivatives are plugged into the infantFS (which requires neurological orientation), it may end up with errors for some cases not the others. This docker image however, can't handle all those variances. In one case with the orientation as aligned with infantFS, I do see the new docker image can tolerate the slight difference between the affine matrix. I'll see if aligning my error cases to the requested orientation can resolve the rest. Thank you!

[xxnagit]

I have tried some more subjects, by orienting them as required, the error would now be reported on the affine matrix not similar enough. This can be manually changed though, but if new version would retain the original affine matrix, that is quite appreciated! Thank you so much!

[mgxd]

@xxnagit can you clarify - are the bibsnet outputs a different orientation than the input data?

[xxnagit]

Sorry to make it confused. The outputs of bibsnet are the same of original inputs. The problem is on my end, my inputs are not of the same orientation which I supposed they were at the beginning.