Research map / Label-efficient and robust learning / Recognition and long-tail
MammoDG: Generalisable Deep Learning Advances Cross-Domain Multi-Center Breast Cancer Screening
Yijun Yang, Shujun Wang, Lihao Liu, Sarah E. Hickman, Fiona J. Gilbert, Carola‐Bibiane Schönlieb, Angelica I. Avilés-Rivero
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Cross-Channel Cross-View Enhancement (CVE) (ablation) | two-branch ResNet18 baseline · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC: +0.0212 AUC, +0.0178 ACC | Improved |
| add Cross-Channel Enhancement (CE) within CVE (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC: +0.0300 AUC, +0.0280 ACC | Improved |
| add Cross-View Enhancement (VE) on top of CE (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC: +0.0050 AUC, +0.0254 ACC | Improved |
| add Mixstyle augmentation (+MS) (ablation) | MammoDG · INBreast (unseen) | AUC (INBreast unseen): 0.7780 -> 0.8193 | Improved |
| add Global Encoder (+GE) (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC: 0.8207 -> 0.8213 | Improved |
| add Multi-Instance Contrastive Learning (MICL) instead of vanilla classifier head (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC: +0.0128 AUC, +0.0227 ACC | Improved |
| add MICL instead of MIL aggregator (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC: +0.0117 AUC, +0.0166 ACC | Improved |
| use equal 1:1:1 weights for CC, MLO and breast-level predictions (ablation) | MammoDG · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC | Improved |
| set contrastive weight lambda to 0.5 (ablation) | MICL · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC | Improved |
| divide each image into 4x4 tiles in MICL (ablation) | MICL · Mammography overall (CBIS, CMMD, INBreast) | AUC and ACC | Improved |
| MammoDG vs best generalisable method MSVCL(FCOS) on unseen domains | MammoDG · Unseen domains overall | AUC and ACC (unseen overall): +0.0386 AUC, +0.0468 ACC | Improved |
| MammoDG vs GMIC on seen domains | MammoDG · Seen domains average | AUC and ACC (seen average): +0.0082 AUC, +0.0070 ACC | Improved |
Closest papers by meaning
- NaFV-Net: An Adversarial Four-view Network for Mammogram Classification (AAAI 2025 · similarity 0.85)
- PersonalizedUS: Interpretable Breast Cancer Risk Assessment with Local Coverage Uncertainty Quantification (AAAI 2025 · similarity 0.83)
- TinyMIG: Transferring Generalization from Vision Foundation Models to Single-Domain Medical Imaging (ICML 2025 · similarity 0.82)
- Are nuclear masks all you need for improved out-of-domain generalisation? A closer look at cancer classification in histopathology (NeurIPS 2024 · similarity 0.81)
- Prompt2Perturb (P2P): Text-Guided Diffusion-Based Adversarial Attack on Breast Ultrasound Images (CVPR 2025 · similarity 0.81)
- Learning Generalized Medical Image Segmentation from Decoupled Feature Queries (AAAI 2024 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
