Research map / Image generation / Fast diffusion sampling
ECHO: Efficient Chest X-ray Report Generation with One-step Block Diffusion
Lifeng Chen, tianqi you, Hao Liu, Zhimin Bao, Jile Jiao, Xiao Han, Zhicai Ou, Tao Sun, Mou XiaoFeng, Xiaojie Jin, Yi Xu
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 step-wise token reweighting (SW) (ablation) | DCD · CheXpert-Plus | PPL: 23.72 -> 21.07 | Improved |
| add step-wise token reweighting (SW) (ablation) | DCD · CheXpert-Plus | RaTEScore: 54.87 -> 56.30 | Improved |
| add eos cross-entropy loss (CE) (ablation) | DCD · CheXpert-Plus | ROUGE-L: 52.44 -> 56.14 | Improved |
| add eos cross-entropy loss (CE) (ablation) | DCD · MIMIC-CXR | CIDEr: 3.65 -> 4.05 | Improved |
| add eos cross-entropy loss (CE) (ablation) | DCD · PPL | Perplexity: 21.07 -> 18.83 | Improved |
| replace forward KL with reverse KL (RKL) (ablation) | DCD · MIMIC-CXR | CIDEr: 3.65 -> 3.48 | Hurt |
| replace forward KL with reverse KL (RKL) (ablation) | DCD · MIMIC-CXR | SemScore: 43.70 -> 42.43 | Hurt |
| add DCD distillation versus native one-step decoding (ablation) | ECHO-Base block4 · CheXpert-Plus | ROUGE-L: 41.39 -> 56.14 | Improved |
| add DCD distillation versus native one-step decoding (ablation) | ECHO-Base block8 · CheXpert-Plus | ROUGE-L: 41.27 -> 55.21 | Improved |
| train on normalized instead of raw reports (Stage I) (ablation) | ECHO-AR · CheXpert-Plus | ROUGE-L: 23.82 -> 56.89 | Improved |
| train on normalized instead of raw reports (Stage III) (ablation) | ECHO block4 · CheXpert-Plus | ROUGE-L: 18.79 -> 56.14 | Improved |
| increase RAD training steps (ablation) | ECHO-Base · TPF during RAD | TPF: 1.62 -> 2.17 | Improved |
Closest papers by meaning
- Enhanced Contrastive Learning with Multi-view Longitudinal Data for Chest X-ray Report Generation (CVPR 2025 · similarity 0.86)
- Addressing Asynchronicity in Clinical Multimodal Fusion via Individualized Chest X-ray Generation (NeurIPS 2024 · similarity 0.86)
- Trade-Offs in Fine-Tuned Diffusion Models between Accuracy and Interpretability (AAAI 2024 · similarity 0.85)
- CXPMRG-Bench: Pre-training and Benchmarking for X-ray Medical Report Generation on CheXpert Plus Dataset (CVPR 2025 · similarity 0.85)
- HC-LLM: Historical-Constrained Large Language Models for Radiology Report Generation (AAAI 2025 · similarity 0.85)
- Think while You Generate: Discrete Diffusion with Planned Denoising (ICLR 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.
