Repository navigation
fix(metrics): count true positives for bool inputs in confusion matrix and F-beta score - #9189
Open
tritsystem wants to merge 2 commits into
Open
tritsystem wants to merge 2 commits into
tritsystem wants to merge 2 commits into
Conversation
Contributor
|
No actionable comments were generated in the recent review.
 |
| Check name | Status | Explanation | Resolution |
|---|---|---|---|
| Docstring Coverage |  |
Docstring coverage is 66.67% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 6 functions across 4 files. | Write docstrings for the functions missing them to satisfy the coverage threshold. |
-
Check name Status Explanation Title check The title clearly summarizes the main fix for Boolean true-positive counting in the confusion matrix and F-beta score. Description check The description explains the defect, impact, cause, fix, and added tests. It is mostly complete, although the issue number is missing and several template checklist items are omitted. Linked Issues check Check skipped because no linked issues were found for this pull request. Out of Scope Changes check Check skipped because no linked issues were found for this pull request.
Finishing Touches
-
🧪 Generate unit tests (beta)
-
- Create a new PR
-
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.
Comment @coderabbitai help to get the list of available commands.
…atrix get_confusion_matrix finds true positives with (y_pred + y) == 2. For bool tensors that sum is a logical OR, which is never 2, so bool masks such as pred > 0.5 lost every true positive: a perfect prediction scored sensitivity 0 and F1 0. Cast bool inputs to float before counting; float and integer inputs are unchanged. Assisted-by: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: tritsystem <gbranaa4@gmail.com>
get_f_beta_score has the same (y_pred + y) == 2 true-positive test as get_confusion_matrix; both lost their float casts in Project-MONAI#7187. For bool masks the sum is a logical OR, so every true positive was dropped and a perfect prediction scored F-beta 0. Cast bool inputs to float before counting, as in the previous commit; float and integer inputs are unchanged. Assisted-by: Claude Opus 5.5 <noreply@anthropic.com> Signed-off-by: tritsystem <gbranaa4@gmail.com>
tritsystem
force-pushed
the
fix/confusion-matrix-bool-inputs
branch
from
October 25, 2026 11:42
ba59ee7 to
67a70ae
Compare
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Files:
monai/metrics/confusion_matrix.py(get_confusion_matrix) andmonai/metrics/f_beta_score.py(get_f_beta_score). Both count true positives with(y_pred + y) == 2, which works for float/int tensors. If bothy_predandyare bool tensors, PyTorch's+does logical OR, not addition, soTrue + TruegivesTrue(1), never 2 -- the true-positive count comes out as zero every time for bool inputs.Impact: a perfect bool prediction (identical bool masks) scores sensitivity 0, precision NaN, F1 0, F-beta 0 -- a perfect match graded as a total failure. The same perfect prediction in float form correctly scores 1.0. This only triggers when both inputs are bool; if either is float, PyTorch promotes the other and it works. Checked every other metric in MONAI (Dice, MeanIoU, GeneralizedDice, Hausdorff, surface distance) -- none have this pattern, so it's isolated to these two functions.
This is a regression: #7187 (merged Nov 2023) removed the
.float()casts from both functions as "unnecessary".Fix: cast bool inputs to float at the top of both functions; float/int inputs are unchanged.
Types of changes
Fixes # .