Research map / Graph learning / Graph self-supervised learning
Act Like a Radiologist: Towards Reliable Multi-View Correspondence Reasoning for Mammogram Mass Detection
Yuhang Liu, Fandong Zhang, Chaoqi Chen, Siwen Wang, Yizhou Wang, Yizhou Yu
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 |
|---|---|---|---|
| use pseudo landmarks instead of uniform grids (ablation) | AG-RCNN · DDSM | R@0.5 recall: 77.7 -> 82.0 | Improved |
| increase pseudo landmark nodes from PL(1, 1) to PL(66, 71) (ablation) | AG-RCNN · DDSM | R@0.5 recall: 77.0 -> 82.0 | Improved |
| increase pseudo landmark nodes from PL(1, 1) to PL(66, 71) (ablation) | AG-RCNN · In-house | R@0.5 recall: 84.5 -> 87.6 | Improved |
| use bipartite graph node mapping instead of cropping a fixed region (ablation) | AG-RCNN PL(66, 71) · DDSM | R@0.5 recall: 78.4 -> 82.0 | Improved |
| use k=3 instead of k=1 with dense nodes (ablation) | AG-RCNN PL(66, 71) · DDSM | R@0.5 recall: 80.1 -> 82.0 | Improved |
| use larger k=3 with sparse nodes instead of k=2 (ablation) | AG-RCNN PL(21, 25) · DDSM | R@0.5 recall: 81.1 -> 80.1 | Hurt |
| add geometric edges H^g (ablation) | Mask RCNN with bipartite graph · DDSM | R@0.5 recall: 76.7 -> 77.7 | Improved |
| add semantic edges H^s (ablation) | Mask RCNN with bipartite graph · DDSM | R@0.5 recall: 76.7 -> 78.4 | Improved |
| combine H^g and H^s (ablation) | Mask RCNN with bipartite graph · DDSM | R@0.5 recall: 76.7 -> 79.5 | Improved |
| use multi-branch IGN(1, 3, 5) instead of IGN(1) (ablation) | Inception graph network · DDSM | R@0.5 recall: 78.0 -> 79.1 | Improved |
| add BGN (ablation) | Mask RCNN · DDSM | R@0.5 recall: 76.7 -> 79.5 | Improved |
| add IGN (ablation) | Mask RCNN · DDSM | R@0.5 recall: 76.7 -> 79.1 | Improved |
Closest papers by meaning
- Cross-View Correspondence Reasoning Based on Bipartite Graph Convolutional Network for Mammogram Mass Detection (CVPR 2020 · similarity 0.93)
- NaFV-Net: An Adversarial Four-view Network for Mammogram Classification (AAAI 2025 · similarity 0.86)
- "Why Is There a Tumor?": Tell Me the Reason, Show Me the Evidence (ICML 2025 · similarity 0.83)
- Multi-view Evidential Learning-based Medical Image Segmentation (AAAI 2025 · similarity 0.83)
- TumorChain: Interleaved Multimodal Chain-of-Thought Reasoning for Traceable Clinical Tumor Analysis (ICLR 2026 · similarity 0.83)
- PathView-Bench: Can Multimodal Large Language Models Achieve Fine-grained Multiscale Understanding of Pathology Images? (NeurIPS 2026 · similarity 0.83)
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.
