AbdomenAtlas 3.0 Mini: Abdominal CT Scans with Radiology Reports and Tumor Annotations

AbdomenAtlas 3.0 Mini is a public dataset containing more than 9,000 high-quality abdominal CT scans paired with radiology reports. It includes per-voxel annotations for tumors in the liver, kidney, and pancreas. The dataset was created by AbdomenAtlas and was last updated on Hugging Face in October 2025.

Use Cases

• Training segmentation models for liver, kidney, and pancreatic tumors based on the per-voxel annotations.
• Developing or benchmarking automated radiology report generation models using the paired CT scans and text reports.
• Conducting research on multi-modal medical AI that links imaging findings with clinical text.

Strengths

• Contains more than 9,000 CT scans, providing a substantial volume of medical imaging data.
• Includes high-quality, per-voxel annotations for tumors in three major abdominal organs.
• Pairs each CT scan with a corresponding radiology report, enabling multi-modal analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

AbdomenAtlas

Freshness

Last updated 2025-10-24 19:50:53; freshness should be verified.

Geography

Regionus (suggests United States)