mlsanityData to 14 Oct 2026

JustLLMGRPO: Radiographic Control for Chest X-Ray Generation

arXiv 2608.08046 ↗First posted 8 Aug 2026cs.AIcited by 0

Not on the map. Why?

Abstract (arXiv)

Text-conditioned chest X-ray generation aims to synthesize realistic radiographs that faithfully depict specified findings. Existing work has primarily improved quality by updating image generators, implicitly treating prompts as fixed after CXR-domain adaptation. We show that this generator-centric view leaves a substantial optimization dimension underexplored. With a CXR-adapted Sana generator frozen, one-pass reformulation by an unmodified LLM reduces RadDINO-FID from 54.225 to 27.572. Prompt analysis shows that the LLM suppresses temporal comparisons, uncertainty, and other non-renderable report content while emphasizing visible radiographic findings. However, unconstrained reformulation reduces BioViL-T alignment with source prompts from 0.695 to 0.609. We therefore introduce JustLLMGRPO, which applies standard Group Relative Policy Optimization (GRPO) only to the LLM prompt policy while keeping Sana frozen. Group-relative radiology-aware image feedback retains visual focus while preserving source-prompt alignment. On CheXGenBench, JustLLMGRPO reduces RadDINO-FID to 26.780, a 50.6% improvement over direct prompting, while maintaining alignment (0.696 versus 0.695). It also achieves state-of-the-art distribution coverage and downstream classification utility. These results show that substantial performance can remain latent in how radiographic information is expressed to an adapted generator. Code is publicly available at https://github.com/pxcai/JustLLMGRPO.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 11 of its 25 references that are in the map. First posted on arXiv 8 Aug 2026.

Builds on

11 of 25 references.

References · 11

  1. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  2. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,423
  3. 2022Large Language Models Are Human-Level Prompt Engineerscited by 1,045
  4. 2023Training Diffusion Models with Reinforcement Learningcited by 982
  5. 2022RLPrompt: Optimizing Discrete Text Prompts with Reinforcement Learningcited by 366
  6. 2023Aligning Text-to-Image Diffusion Models with Reward Backpropagationcited by 288
  7. 2024Exploring scalable medical image encoders beyond text supervisioncited by 132
  8. 2024Improving Text-to-Image Consistency via Automatic Prompt Optimizationcited by 90
  9. 2023Cascaded Latent Diffusion Models for High-Resolution Chest X-ray Synthesiscited by 32
  10. 2023RadEdit: stress-testing biomedical vision models via diffusion image editingcited by 28
  11. 2025CheXGenBench: A Unified Benchmark For Fidelity, Privacy and Utility of Synthetic Chest Radiographscited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026CXRMate-2: Structured Multimodal Temporal Embeddings and Tractable Reinforcement Learning for Clinically Acceptable Chest X-ray Radiology Report Generation
  2. 2026REVA-PO: Stabilizing Reinforcement Learning for Chest X-ray Report Generation
  3. 2026DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy Optimization
  4. 2024Advancing Text-Driven Chest X-Ray Generation with Policy-Based Reinforcement Learning
  5. 2024CheXalign: Preference fine-tuning in chest X-ray interpretation models without human feedback
  6. 2026UniX: Unifying Autoregression and Diffusion for Chest X-Ray Understanding and Generation
  7. 2026CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
  8. 2025LLaMA-XR: A Novel Framework for Radiology Report Generation using LLaMA and QLoRA Fine Tuning