IU-Xray-RRG: Radiology Report Generation Subsets

IU-Xray-RRG packages two evaluation-focused slices from the IU X-Ray collection for radiology report generation experiments. Each configuration pairs a single-view chest radiograph (frontal) with the corresponding section of the diagnostic report, mirroring the benchmarks used in the CCD studies.

📚 Dataset Overview

• Designed as a test-only suite for benchmarking and zero-shot evaluation.
• Matches the protocol adopted by CCD and Libra when assessing report-generation quality.

---

🧾 Data Format

Each record contains:

• main_image: a chest X-ray stored as a PIL Image object.
• Section text: findings_section or impression_section, depending on the chosen configuration.
• default_prompt: a lightweight prompt scaffold for generation models.

🚀 How to Use

from datasets import load_dataset

# Load a specific subset (e.g., findings_section)
ds = load_dataset("X-iZhang/IU-Xray-RRG", name="findings_section", split="test")

# Display an image
from PIL import Image
ds[0]["main_image"].show()

# View sample
print(ds[0]["findings_section"])

✏️ Citation

@article{zhang2025ccd,
  title={CCD: Mitigating Hallucinations in Radiology MLLMs via Clinical Contrastive Decoding},
  author={Zhang, Xi and Meng, Zaiqiao and Lever, Jake and Ho, Edmond SL},
  journal={arXiv preprint arXiv:2509.23379},
  year={2025}
}