Semi-Supervised Deep Learning-Based Model for Segmentation of Breast Arterial Calcification on Screening Mammograms
Bibliographic record
Abstract
PURPOSE: Breast arterial calcification (BAC), detectable on routine mammograms, offers a promising independent risk factor for cardiovascular disease (CVD) risk stratification. However, current BAC assessment methods lack standardization and rely on subjective interpretations. This study introduces a semi-supervised deep learning (DL) model to automate BAC severity grading, enhance cross-system generalizability, and align with clinical consensus. METHODS: A U-Net-based segmentation model was trained on 2560 annotated screening mammograms from 7 vendors. A semi-supervised learning strategy employing progressive pseudo-labeling incorporated 6000 unlabeled images to enhance model robustness. BAC severity was graded by thresholding the percentage area covered by BAC and benchmarked against radiologists' assessments using Canadian Society of Breast Imaging (CSBI) guidelines. Performance was evaluated using the Jaccard Similarity Coefficient (JSC) for segmentation, along with accuracy, precision, F1-score, and recall. For detecting clinically significant (Grade 3) BAC, sensitivity, specificity, and area under the curve (AUC) were assessed. Agreement with experts was evaluated using weighted kappa statistics. RESULTS: The proposed model achieved a JSC of 0.614, an accuracy of 0.991, an F1-score of 0.756, a precision of 0.763, and a recall of 0.764. It demonstrated superior segmentation accuracy compared to the baseline U-Net model. Agreement with consensus radiologists was high, with a weighted kappa of 0.90, 95% CI = (0.70, 1.00). For clinically significant (Grade 3) BAC, the model achieved an AUC of 0.87, 95% CI = (0.72, 1.00), sensitivity of 0.80, and specificity of 0.93. CONCLUSION: The framework holds promise for clinical adoption, integrating into mammography workflows and improving women's cardiovascular risk stratification.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".