Enhancing Clinical Interpretability in BI-RADS Classification: Segmentation-Assisted Approach Using U-Net and ResNet50 for Mammography
Bibliographic record
Abstract
Breast cancer is the most prevalent cancer among women worldwide, requiring accurate mammographic assessment for early diagnosis. Significant inter-observer variability exists in differentiating BI-RADS 3 (probably benign) from BI-RADS 4 and 5 (suspicious and malignant). Traditional deep learning methods using end-to-end classification on full mammograms lack interpretability, reducing clinical confidence. We propose a two-stage framework separating lesion segmentation and classification to enhance transparency. U-Net achieves validation Dice coefficient of 0.9771 for lesion localization with strong boundary definition on expert-annotated ground truth. ResNet50 then classifies segmented regions into BI-RADS categories. Testing on CBIS-DDSM yields test accuracy of 0.7197 with recall of 0.91 for BI-RADS 4 and 5, effectively detecting clinically critical abnormalities. However, BI-RADS 3 recall is 0.27, reflecting inherent classification confusion at the diagnostic boundary where expert radiologists show only moderate agreement (kappa 0.47-0.71). The segmentation-first approach outperforms baseline full-image classification (accuracy 0.70) by 0.02 and provides interpretable masks enabling radiologists to verify decision-contributing regions. This explicit separation enhances interpretability and facilitates human-AI collaboration in breast cancer screening.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.011 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".