A deep learning framework for accurate mammographic mass classification using local context attention module
Bibliographic record
Abstract
BACKGROUND: Dense breast tissue significantly increases breast cancer (BC) risk. However, current mammographic methods for classifying BC are often subjective and unreliable, which complicates the task of accurate evaluation. PURPOSE: This study introduces a deep learning method with a local context attention module (LCAM), using dual mammogram views aligned with BI-RADS to enhance grading consistency and accuracy in BC classification across four groups by leveraging local context around masses. METHODS: Specific regions of interest (ROIs) containing dense tissue around breast masses are identified from dual mammogram views, providing additional insights for predicting BC BI-RADS categories. These ROIs are then input into a convolutional neural network (CNN)-based model, which is crucial for selecting and differentiating radiomic features associated with BI-RADS. To enhance our model's ability to distinguish salient radiomic features associated with mass malignancy, the LCAM sequentially infers attention maps along two separate dimensions: channel and spatial. These attention maps are subsequently multiplied with the input feature map for adaptive feature refinement. RESULTS: Examining 3020 patients across four BI-RADS categories while leveraging dual mammogram views demonstrates the robust performance of the proposed framework, achieving a sensitivity of 82.46% and a specificity of 91.42% in identifying BI-RADS grading relevant to breast masses. CONCLUSIONS: We introduced a novel CNN-based framework that utilizes dual mammogram views for the BC classification. It utilizes LCAM, which further understands the local characteristics surrounding breast masses, aiming to enhance the accuracy and consistency of classification outcomes.
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 | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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 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".