Research map / Label-efficient and robust learning / Self-supervised and contrastive
Advancing Radiograph Representation Learning with Masked Record Modeling
Hong-Yu Zhou, Chenyu Lian, Liansheng 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 |
|---|---|---|---|
| combine masked report modeling and masked radiograph modeling with super-resolution restoration (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 69.9 -> 79.4; 75.2 -> 84.0; 80.3 -> 85.9 | Improved |
| add masked report modeling (L_R) (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 74.7 -> 79.4; 81.3 -> 84.0; 85.1 -> 85.9 | Improved |
| add masked radiograph modeling (L_I) (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 76.7 -> 79.4; 82.2 -> 84.0; 84.7 -> 85.9 | Improved |
| add super-resolution restoration (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 78.8 -> 79.4; 83.7 -> 84.0; 85.7 -> 85.9 | Improved |
| add hybrid multimodal features to image restoration (ablation) | MRM · NIH ChestX-ray | AUC: 79.4 vs. 78.9; 84.0 vs. 83.6; 85.9 vs. 85.7 | No clear effect |
| use global average pooling instead of global maximum pooling (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 79.4 vs. 77.2; 84.0 vs. 83.8; 85.9 vs. 86.0 | Mixed |
| set radiograph masking ratio to 75% rather than 50% (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 79.4 vs. 78.8; 84.0 vs. 84.4; 85.9 vs. 86.1 | Mixed |
| set loss weight lambda to 1 rather than 2 or 0.5 (ablation) | MRM · NIH ChestX-ray 1%, 10%, 100% | AUC: 79.4 (lambda=1); 78.6 (lambda=2); 79.0 (lambda=0.5); 84.0; 83.5; 83.7; 85.9; 85.4; 85.9 | Mixed |
| use MRM pre-training rather than REFERS pre-training | radiograph representation learning · CheXpert 1% | AUC: 88.5% versus previous counterparts with 100% annotations | Improved |
| use MRM pre-training rather than REFERS | radiograph representation learning · NIH ChestX-ray 1%, 10% | average AUC: 2.7%; 3.1% | Improved |
| use MRM rather than self-supervised and report-supervised pre-training | pneumonia classification · Viral vs. Bacterial | score: 91.5 vs. 80.4 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GLoRIA* — GLoRIA: A Multimodal Global-Local Representation Learning Framework for Label-Efficient Medical Image Recognition (ICCV 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Bootstrapping Chest CT Image Understanding by Distilling Knowledge from X-ray Expert Models (CVPR 2024 · similarity 0.86)
- A World Model of Radiologist Reading for Medical Image Representation Learning (NeurIPS 2026 · similarity 0.84)
- Autoregressive Sequence Modeling for 3D Medical Image Representation (AAAI 2025 · similarity 0.84)
- CheXWorld: Exploring Image World Modeling for Radiograph Representation Learning (CVPR 2025 · similarity 0.84)
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.84)
- Large-scale and Fine-grained Vision-language Pre-training for Enhanced CT Image Understanding (ICLR 2025 · similarity 0.84)
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.
