5,195 CT volumes featuring 9 annotated organ classes form the basis of AbdomenAtlas 1.0 for multi-organ segmentation. Developed by MrGiovanni and presented at NeurIPS 2023, it serves as a large-scale resource for abdominal medical imaging.
Use Cases
- Training deep learning models for semantic segmentation of 9 abdominal organ classes
- Evaluating the generalizability of segmentation algorithms across 5,195 diverse CT volumes
- Developing automated radiology workflows for identifying anatomical structures in abdominal scans
Strengths
- 5,195 CT volumes
- 9 annotated organ classes
- Peer-reviewed at NeurIPS 2023
Limitations
- Restricted to abdominal anatomy
- Unknown metadata regarding scanner types or patient demographics
Provenance
- Source
- MrGiovanni (NeurIPS 2023)
- Collection Method
- Annotated CT volumes
- Freshness
- Last updated November 2025