Research map / 3D vision / 3D generation
X-WIN: Building Chest Radiograph World Model via Predictive Sensing
Zefan Yang, Ge Wang, James Hendler, Mannudeep K. Kalra, Pingkun Yan
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 direct yaw rotation instead of step-wise rotation as action (ablation) | X-WIN · VinDr linear probing | AUROC: 0.857 -> 0.925 | Improved |
| use yaw-only rotation instead of yaw, pitch and roll (ablation) | X-WIN · VinDr linear probing | AUROC: 0.871 -> 0.925 | Improved |
| add structure-preserving domain adaptation loss (ablation) | X-WIN · sim-to-real representation similarity | cosine similarity: 0.845 -> 0.967 | Improved |
| rotation step size 3 degrees instead of 5-15 degrees (ablation) | X-WIN · VinDr linear probing | AUROC | Improved |
| rotation step size 3 degrees instead of 1 degree (ablation) | X-WIN · VinDr linear probing | AUROC | Improved |
| combine InfoNCE and MIM losses (ablation) | X-WIN · VinDr linear probing | AUROC | Improved |
| reduce any loss weight to 0.0 (ablation) | X-WIN · VinDr linear probing | AUROC | Hurt |
| distill 3D CT knowledge (X-WIN) vs 2D CheXWorld | X-WIN ViT-Base · 5 CXR benchmarks avg | AUROC: 0.844 -> 0.883 | Improved |
| X-WIN vs Ark+ on RSNA/JSRT | X-WIN ViT-Base · VinDr | AUROC: 0.906 -> 0.905 | No clear effect |
| X-WIN vs Ark+ on 4-shot COVIDx | X-WIN · COVIDx 4-shot | AUROC: 0.872 -> 0.868 | Hurt |
| X-WIN vs Ark+ with all labels | X-WIN · COVIDx all | AUROC: 0.985 -> 0.993 | Improved |
Closest papers by meaning
- CXReasonBench: A Benchmark for Evaluating Structured Diagnostic Reasoning in Chest X-rays (NeurIPS 2025 · similarity 0.88)
- Bootstrapping Chest CT Image Understanding by Distilling Knowledge from X-ray Expert Models (CVPR 2024 · similarity 0.86)
- BenchX: A Unified Benchmark Framework for Medical Vision-Language Pretraining on Chest X-Rays (NeurIPS 2024 · similarity 0.86)
- Rethinking Visual Attribution for Chest X-ray Reasoning in Large Vision Language Models (NeurIPS 2026 · similarity 0.86)
- CheXWorld: Exploring Image World Modeling for Radiograph Representation Learning (CVPR 2025 · similarity 0.86)
- AOR: Anatomical Ontology-Guided Reasoning for Medical Large Multimodal Model in Chest X-Ray Interpretation (NeurIPS 2025 · similarity 0.85)
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.
