AbdomenAtlas1.0Mini: 5,195 Annotated Abdominal CT Volumes

5,195 fully-annotated abdominal CT volumes constitute one of the largest datasets of its kind. The dataset includes annotations for eight anatomical structures: spleen, liver, kidneys, stomach, gallbladder, pancreas, aorta, and IVC. It was created by the CCVL research group at Johns Hopkins University and last updated on January 16, 2025.

Use Cases

• Benchmarking semantic segmentation models based on the eight annotated abdominal organs and vessels.
• Training medical image analysis models for organ localization and segmentation.
• Developing pre-training algorithms for 3D medical imaging tasks using a large-scale annotated dataset.
• Comparing algorithm performance as part of the AbdomenAtlas Benchmarking Project (Touchstone).

Strengths

• One of the largest annotated CT datasets with 5,195 volumes.
• Includes annotations for eight specific anatomical structures (spleen, liver, kidneys, stomach, gallbladder, pancreas, aorta, IVC).
• Created by the CCVL research group at Johns Hopkins University, a reputable institution.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or institutional bias inherent to the source collection.

Provenance

Source

CCVL research group at Johns Hopkins University

Collection Method

Collection and annotation of CT volumes, likely from clinical sources.

Freshness

Last updated 2025-01-16 18:58:06