✨♻️ refactor preprocessing

[dyollb]

What do these changes do?

The PR adds

• SelectChannel transform
• an option to crop the foreground using an otsu mask (thresholding the first channel)
• various small fixes
• improve predict command, to write metrics in csv file, plot summary confusion matrix

Related issue/s

How to test

• run pytest

Checklist

• Non-breaking change (fix or new feature that would not break existing functionality).
• Unit tests for the changes and run locally with the command pytest tests
• Runs on minimum Python version

[codecov]

Codecov Report

Attention: Patch coverage is 60.75949% with 62 lines in your changes are missing coverage. Please review.

Project coverage is 56.77%. Comparing base (288663f) to head (dfd3452).

Files	Patch %	Lines
src/segmantic/seg/monai_unet.py	16.07%	45 Missing and 2 partials ⚠️
src/segmantic/seg/losses.py	73.17%	5 Missing and 6 partials ⚠️
src/segmantic/seg/transforms.py	92.00%	1 Missing and 1 partial ⚠️
src/segmantic/seg/visualization.py	0.00%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #64      +/-   ##
==========================================
+ Coverage   55.47%   56.77%   +1.30%     
==========================================
  Files          21       23       +2     
  Lines        1325     1409      +84     
  Branches      237      251      +14     
==========================================
+ Hits          735      800      +65     
- Misses        550      562      +12     
- Partials       40       47       +7     

Flag	Coverage Δ
unittests	56.77% <60.75%> (+1.30%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[dyollb]

gamma=2.0

[dyollb]

this should be AsymmetricFocalLoss:

see https://github.com/mlyg/unified-focal-loss/blob/411c9c5ce43b2ef847f0903d1b841512ad8d2eee/loss_functions.py#L193

to compute AsymmetricUnifiedFocalLoss we also need the AsymmetricTverskyLoss

see https://github.com/mlyg/unified-focal-loss/blob/411c9c5ce43b2ef847f0903d1b841512ad8d2eee/loss_functions.py#L351