RadCases: ACR Appropriateness Criteria Labels for Radiology Case Summaries

A dataset containing labels for patient case summaries according to the American College of Radiology (ACR) Appropriateness Criteria. The original case summaries are not included due to access restrictions; instead, the dataset provides cryptographically secure SHA-512 hashes of those summaries. The dataset was created by michaelsyao and last updated on March 1, 2025.

Use Cases

• Training models to classify radiology case appropriateness based on ACR criteria labels.
• Developing anonymization or hashing techniques for sensitive clinical text data.
• Studying patterns in radiology referral or diagnostic decision-making.
• Benchmarking NLP models on tasks involving structured medical guidelines.

Strengths

• Labels are structured according to the authoritative ACR Appropriateness Criteria.
• Patient privacy is protected via SHA-512 hashing of original case summaries.
• Dataset was updated on March 1, 2025, indicating recent maintenance.

Limitations

• The primary data (patient case summaries) are not publicly accessible, limiting direct analysis.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.

Provenance

Source

HuggingFace dataset created by michaelsyao.

Collection Method

Constructed from sources requiring credentialed access; public release contains derived labels and hashes.

Time Range

null

Freshness

Last updated 2025-03-01 18:58:39

Geography

null