RSNA Intracranial Hemorrhage Detection: 25,000+ Labeled Head CT Scans

RSNA assembled this dataset in 2019 for the RSNA Intracranial Hemorrhage Detection AI Challenge. It consists of de-identified head CT studies provided by four research institutions. Over 60 volunteer expert radiologists labeled more than 25,000 exams for the presence and subtype of acute intracranial hemorrhage.

Use Cases

• Train image classification models to detect intracranial hemorrhage based on labeled CT scans.
• Develop multi-label classifiers for hemorrhage subtype classification using expert-provided labels.
• Benchmark AI model performance against a standardized challenge dataset.
• Research computer vision techniques for 3D medical imaging analysis.

Strengths

• Over 25,000 labeled medical imaging exams provide substantial scale for model training.
• Labels were created by a group of over 60 volunteer expert radiologists, suggesting high annotation quality.
• Data was assembled for a formal AI challenge, implying a standardized and curated structure.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• Row count is unknown, which may limit suitability assessment for specific model architectures.

Provenance

Source

Radiological Society of North America (RSNA)

Collection Method

De-identified head CT studies were provided by four research institutions and labeled by over 60 volunteer expert radiologists.

Time Range

2019

Freshness

Dataset was assembled in 2019; last update date is unknown.

Geography

null