Research map / Vision-language and multimodal / Human and embodied modeling
METransformer: Radiology Report Generation by Transformer With Multiple Learnable Expert Tokens
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 bilinear-attention encoder (ablation) | Baseline report-generation model · IU-Xray | CIDEr: 0.337 -> 0.346 | Helped |
| add bilinear-attention encoder (ablation) | Baseline report-generation model · MIMIC-CXR | CIDEr: 0.275 -> 0.287 | Helped |
| add expert tokens with the bilinear-attention encoder (ablation) | Baseline report-generation model · IU-Xray | CIDEr: 0.346 -> 0.402 | Helped |
| add expert tokens with the bilinear-attention encoder (ablation) | Baseline report-generation model · MIMIC-CXR | CIDEr: 0.287 -> 0.334 | Helped |
| add orthogonal loss to expert tokens (ablation) | METransformer with expert tokens · IU-Xray | CIDEr: 0.402 -> 0.426 | Helped |
| add orthogonal loss to expert tokens (ablation) | METransformer with expert tokens · MIMIC-CXR | CIDEr: 0.334 -> 0.352 | Helped |
| select the report by expert voting instead of averaging expert word probabilities (ablation) | METransformer with expert tokens and orthogonal loss · IU-Xray | CIDEr: 0.426 -> 0.435 | Helped |
| select the report by expert voting instead of averaging expert word probabilities (ablation) | METransformer with expert tokens and orthogonal loss · MIMIC-CXR | CIDEr: 0.352 -> 0.362 | Helped |
| use seven expert tokens rather than one (ablation) | METransformer · IU-Xray | CIDEr: 0.346 -> 0.435 | Helped |
Closest papers by meaning
- Instance-level Expert Knowledge and Aggregate Discriminative Attention for Radiology Report Generation (CVPR 2024 · similarity 0.86)
- Bootstrapping Large Language Models for Radiology Report Generation (AAAI 2024 · similarity 0.85)
- LLM-RG4: Flexible and Factual Radiology Report Generation Across Diverse Input Contexts (AAAI 2025 · similarity 0.85)
- AHIVE: Anatomy-aware Hierarchical Vision Encoding for Interactive Radiology Report Retrieval (CVPR 2024 · similarity 0.84)
- Bootstrapping Chest CT Image Understanding by Distilling Knowledge from X-ray Expert Models (CVPR 2024 · similarity 0.84)
- PairAug: What Can Augmented Image-Text Pairs Do for Radiology? (CVPR 2024 · 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.