Hybrid deep learning and active contour approach for enhanced breast lesion segmentation and classification in mammograms
Bibliographic record
Abstract
Accurate segmentation and classification of breast lesions in mammography images are crucial steps in effective breast cancer screening and diagnosis. This study presents a hybrid deep learning and active contour approach to automated mammogram analysis. The proposed methodology leverages the powerful feature extraction capabilities of deep convolutional neural networks and the precise boundary delineation of active contour models. A U-Net is trained on a large dataset of mammogram images to learn discriminative features and generate initial segmentation masks for breast lesions. Subsequently, an active contour refinement stage is employed to fine-tune the segmentation boundaries and enhance lesion delineation accuracy. This integration of active contour models (ACM) with deep learning techniques overcomes traditional image segmentation limitations. Morphological operations and energy minimization techniques are applied to the initial segmentation mask, resulting in highly accurate and refined lesion segmentation. This study investigates the synergistic integration of deep learning with Adaptive Contour Modeling for breast lesion segmentation. Our proposed U-Net_ACM model leverages the strengths of both approaches, demonstrating state-of-the-art performance and outperforming methods relying solely on deep learning or traditional image processing techniques. Evaluation on a test set reveals a 97.34 % accuracy, a Dice coefficient of 0.813, and an Intersection over Union of 0.891 for the U-Net_ACM model. These results surpass the performance of established pre-trained deep learning models such as VGG16, VGG19, and DeepLabV3, highlighting the benefits of the combined approach. This hybrid methodology offers a robust, automated solution for mammogram analysis, potentially improving breast cancer screening outcomes. The superior segmentation quality and overall performance demonstrated by the U-Net_ACM model suggest its potential for enhancing breast cancer screening and diagnosis in clinical settings.
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 | 0.002 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 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".