Add options to segment_3d_image for removing objects touching borders

[lguerard]

Fix #138

[codecov]

Codecov Report

❌ Patch coverage is 0% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 25%. Comparing base (7cb7b68) to head (0eab3dd).
⚠️ Report is 28 commits behind head on devel.

Files with missing lines	Patch %	Lines
src/imcflibs/imagej/objects3d.py	0%	4 Missing ⚠️

Additional details and impacted files

@@         Coverage Diff          @@
##           devel   #139   +/-   ##
====================================
  Coverage     25%    25%           
====================================
  Files         25     25           
  Lines       1771   1763    -8     
====================================
+ Hits         437    443    +6     
+ Misses      1334   1320   -14     

☔ 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.

[ehrenfeu]

This won't have an effect if the non-z parameter is false, as far as I can see.

Either document this or automatically set the non-z to true if the z one is true inside the function. I prefer the latter approach as it seems less surprising.

[lguerard]

I checked this morning and there isn't a function in the 3DImageJSuite that could filter out only objects in Z but not XY.

There is a method in MorpholibJ that I could use though, should I switch to that ?

[ehrenfeu]

Let me clarify what I'm trying to say.

From reading the docstring, what would an innocent developer think 🤔 will happen with this call?

label_imp = segment_3d_image(binary_imp, remove_touching_borders_z=True)

I believe the expectation is that at least some filtering would be happening, which is not the case given the current code execution path.

I don't know if we need a function that purely filters objects in Z, but if the Z-filtering is requested explicitly (without the XY-filtering), it will be silently ignored. My suggestion is to implicitly enable XY-filtering if Z-filtering is requested and to explain this in the docstring.

Would you think we should rather go for the MorphoLibJ approach that allows to only filter in Z? If you foresee any use-cases for this, it's clearly the better approach.

[lguerard]

Yes, I agree.
At the moment, using only the Image3DSuite, there is no way to only filter in Z. If the user wants to do that, then it has to also filter in XY. I'm not happy about this, users might want to filter objects that aren't fully imaged in a stack and remove objects that are in slide 1 or slide max.

I'll improve this using MorpholibJ