Allow custom filters in TissueMask #377

Christopher22 · 2022-01-25T14:22:43Z

Description

This small patch allows a custom "pipeline" of image filters within the TissueMask to optimize for specific demands regarding tissue. The default behavior is not changed.

Types of Changes

Issues Fixed or Closed by This PR

Fixes: Allow custom filters for TissueMask #375

Checklist

My code follows the code style of this project.
My change requires a change to the documentation.
I have updated the documentation accordingly.
I have read the CONTRIBUTING document.
I have added tests to cover my changes.
I have tested the changes and verified that they work and don't break anything (as well as I can manage).

codecov · 2022-01-26T08:24:14Z

Codecov Report

Merging #377 (ecf8a33) into master (df69137) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #377   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           17        17           
  Lines         1411      1430   +19     
  Branches       218       222    +4     
=========================================
+ Hits          1411      1430   +19

Impacted Files	Coverage Δ
histolab/filters/compositions.py	`100.00% <100.00%> (ø)`
histolab/masks.py	`100.00% <100.00%> (ø)`
histolab/tile.py	`100.00% <100.00%> (ø)`

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 df69137...ecf8a33. Read the comment docs.

ernestoarbitrio

Thanks for your contribution @Christopher22 ❤️ !
I have some concerns on this PR 🤷🏽‍♂️ . The first thing is the tests you provided. Please read my comment.
Furthermore @alessiamarcolini can give you other consideration on you work.

tests/unit/test_masks.py

ernestoarbitrio

@Christopher22 I did this CR round with @alessiamarcolini.
Sorry for the bunch of request changes ... but overall looks neat.
Could you please address the comments in order to make us able to do a final round?
Thanks again for you effort 💪🏽 ... we really appreciated it 😄

histolab/filters/compositions.py

ernestoarbitrio · 2022-02-02T18:48:30Z

tests/unit/test_masks.py

+        tissue_mask = TissueMask(*custom_filters)
+        _ = tissue_mask(slide)
+
+        set_custom_filters_.assert_called_once()


I'm not a huge fan of assert_called_once. The reason is that, yes, you're asserting that a proper method has been called, but the interesting info would be also know with which parameters? I'd suggest to switch to assert_called_once_with and check the calling args 😉

I try to remove all of the things. However, I sometimes have no access to the "self" parameter. In this case, I try to check only the remaining arguments.

Sorry I do not understand this point. Could you elaborate? Thx

tests/unit/test_masks.py

ernestoarbitrio · 2022-02-02T18:48:58Z

tests/unit/test_masks.py

+        # and the custom filter itself is called
+        set_custom_filters_.assert_called_once()
+        assert set_custom_filters_.call_args[0][1] == custom_filters[0]
+        custom_filter_call_.assert_called_once()


Regarding the whole test: I feel that testing that RgbToGrayscale is called is not a good strategy, because:

usually we want to mock all the methods/instances that get called/used directly by the method we are testing, so in this case it would be Tile.calculate_tissue_mask

as a proof for my first point: this test is not really checking that the mask is built with the custom filters, because RgbToGrayscale is actually the first filter in the "default" filter sequence (for both Slide and Tile). So this test would actually pass with both the default and the "custom" filters represented by [RgbToGrayscale()] 🤓

Please let me know if this makes sense for you!

@Christopher22 I think you misunderstood my points here. This updated test covers point number 2 but not number 1, the mocked object is not directly called/used by the method you're testing.

Christopher22 · 2022-02-03T13:19:45Z

Thank you so much for your comments! I hope, I was able to fix the things up you mentioned.

As a small refactor, I removed my 'set_custom_filters' and replaced it with a meaningful constructor which fits better in your existing codebase.

ernestoarbitrio

Hey @Christopher22 thanks for all your commitment. I left a couple of furhter minor comments. Could you please take a look? After that we are probably ready to go.

Heads up to @alessiamarcolini ☝🏽

histolab/masks.py

tests/unit/test_masks.py

Christopher22 · 2022-02-08T15:16:51Z

Amazing! Your highlighed areas should be fixed by now :)

alessiamarcolini

Apart from a couple of docstrings improvements, I had to reason carefully about two unit tests (see comments) and please try to cover all of your new code with tests (follow codecov comments), as we cannot merge this PR if it does not provide 100% coverage ✌🏼

histolab/filters/compositions.py

histolab/masks.py

alessiamarcolini · 2022-02-08T19:12:46Z

tests/unit/test_masks.py

            [False, True],
        ]
        tissue_mask_tile_.return_value = expected_mask



not your responsibility, but since you touched this method... this line should also be removed

tests/unit/test_masks.py

alessiamarcolini · 2022-02-08T19:34:10Z

tests/unit/test_masks.py

+        # and the custom filter itself is called
+        set_custom_filters_.assert_called_once()
+        assert set_custom_filters_.call_args[0][1] == custom_filters[0]
+        custom_filter_call_.assert_called_once()


Regarding the whole test: I feel that testing that RgbToGrayscale is called is not a good strategy, because:

usually we want to mock all the methods/instances that get called/used directly by the method we are testing, so in this case it would be Tile.calculate_tissue_mask

as a proof for my first point: this test is not really checking that the mask is built with the custom filters, because RgbToGrayscale is actually the first filter in the "default" filter sequence (for both Slide and Tile). So this test would actually pass with both the default and the "custom" filters represented by [RgbToGrayscale()] 🤓

Please let me know if this makes sense for you!

Christopher22 · 2022-02-10T09:57:08Z

Such cute apes! ;) Your comments should be integrated by now!

alessiamarcolini

Please check the comment I left about the it_knows_its_mask_tile_and_supports_custom_filters test.

PS please squash the commits at the end of the review, otherwise it's more difficult for us to be on track with the changes between review rounds ✌🏼

Christopher22 · 2022-02-21T10:02:23Z

@alessiamarcolini Do you prefer a test such as this instead? :)

alessiamarcolini

Just a minor comment for consistency, the rest is perfect!! 👏🏼

tests/unit/test_masks.py

Christopher22 · 2022-02-25T07:46:10Z

Ready for squash ;)

alessiamarcolini

Let's merge it!

alessiamarcolini · 2022-02-25T13:31:07Z

@Christopher22 thank you so much for your valuable contribution! 👏🏼👏🏼

Now, to have a consistent code base, we would like to provide the same flexibility for the BiggestTissueBoxMask as well. Please let us know if you are interested in contributing again in this regard! Otherwise, we can take care of it, no pressure at all 😉

Christopher22 force-pushed the custom-filters branch from e1fc845 to 2a61e23 Compare January 25, 2022 15:09

alessiamarcolini requested review from alessiamarcolini and ernestoarbitrio January 27, 2022 15:56

ernestoarbitrio requested changes Jan 27, 2022

View reviewed changes

tests/unit/test_masks.py Outdated Show resolved Hide resolved

Christopher22 force-pushed the custom-filters branch from 2a61e23 to 6e10639 Compare January 31, 2022 07:04

ernestoarbitrio requested changes Feb 2, 2022

View reviewed changes

Christopher22 force-pushed the custom-filters branch from 6e10639 to f404b38 Compare February 3, 2022 13:16

ernestoarbitrio requested changes Feb 8, 2022

View reviewed changes

histolab/masks.py Outdated Show resolved Hide resolved

tests/unit/test_masks.py Outdated Show resolved Hide resolved

Christopher22 force-pushed the custom-filters branch from f404b38 to 64be2da Compare February 8, 2022 15:16

alessiamarcolini requested changes Feb 8, 2022

View reviewed changes

Christopher22 force-pushed the custom-filters branch from 64be2da to 11cd78d Compare February 10, 2022 09:56

alessiamarcolini requested changes Feb 20, 2022

View reviewed changes

ernestoarbitrio approved these changes Feb 24, 2022

View reviewed changes

alessiamarcolini requested changes Feb 24, 2022

View reviewed changes

tests/unit/test_masks.py Outdated Show resolved Hide resolved

tests/unit/test_masks.py Outdated Show resolved Hide resolved

alessiamarcolini approved these changes Feb 25, 2022

View reviewed changes

Christopher22 and others added 3 commits February 25, 2022 12:31

Allow custom filters in TissueMask

694c548

Check "calculate_tissue_mask" directly

88e0a4a

Last polishing

41aa5e2

ernestoarbitrio force-pushed the custom-filters branch from d595284 to 41aa5e2 Compare February 25, 2022 11:31

remove meaningless else

ecf8a33

ernestoarbitrio merged commit 34389aa into histolab:master Feb 25, 2022

alessiamarcolini added this to the 0.5.0 milestone Mar 8, 2022

alessiamarcolini added the enhancement New feature or request label Mar 8, 2022

alessiamarcolini mentioned this pull request Mar 8, 2022

Allow custom filters in BiggestTissueBoxMask #387

Merged

11 tasks

Uh oh!

Allow custom filters in TissueMask #377

Allow custom filters in TissueMask #377

Uh oh!

Conversation

Christopher22 commented Jan 25, 2022

Description

Types of Changes

Issues Fixed or Closed by This PR

Checklist

Uh oh!

codecov bot commented Jan 26, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

ernestoarbitrio left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

ernestoarbitrio left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Christopher22 commented Feb 3, 2022

Uh oh!

ernestoarbitrio left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Christopher22 commented Feb 8, 2022

Uh oh!

alessiamarcolini left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Christopher22 commented Feb 10, 2022

Uh oh!

alessiamarcolini left a comment

Choose a reason for hiding this comment

Uh oh!

Christopher22 commented Feb 21, 2022

Uh oh!

alessiamarcolini left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Christopher22 commented Feb 25, 2022

Uh oh!

codecov bot commented Jan 26, 2022 •

edited

Loading

ernestoarbitrio left a comment •

edited

Loading