MIMIC-CXR-RRG: Radiology Report Generation Subsets

A dataset from the MIMIC-CXR collection provides two subsets for Radiology Report Generation tasks. It contains image-report pairs focused on the Findings and Impression sections for frontal-view chest X-rays. The dataset was created by Yamini-1628 and was last updated on December 15, 2025.

Use Cases

• Train models for automated radiology report generation based on frontal chest X-ray images.
• Benchmark natural language generation models using the structured Findings and Impression sections.
• Develop multi-modal AI systems that link medical images with textual clinical impressions.

Strengths

• The dataset is specifically filtered and structured for the Radiology Report Generation task.
• It contains at least 2,461 samples in its 'findings_section' test split.
• It focuses on a specific view type (frontal) and report sections (Findings/Impression), reducing noise.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• The total number of samples, file formats, and license information are unknown.

Provenance

Source

Subsets derived from the MIMIC-CXR dataset.

Collection Method

Carefully filtered and structured from the original source.

Freshness

Last updated 2025-12-15 19:00:52; freshness should be verified.