CADS: 22,022 Whole-Body CT Volumes with 167 Anatomical Segmentations

CADS contains 22,022 CT volumes with automated annotations for 167 anatomical structures ranging from the head to the knee. Created by mrmrx and documented in Arxiv 250722953, it provides 3D segmentation data across multiple anatomical systems.

Use Cases

• Training 3D segmentation models to identify specific organs or bones using the 167 anatomical labels
• Benchmarking automated anatomical landmark detection across the head-to-knee span of the CT volumes
• Developing multi-organ diagnostic frameworks using the 22,022 annotated records

Strengths

• 22,022 CT volumes for high-capacity model training
• 167 anatomical structure labels for granular segmentation
• Head-to-knee anatomical coverage in 3D modality

Limitations

• Labels are generated via a fully automated framework, which may contain machine-generated noise compared to manual expert review
• Geographic bias toward the US region as indicated by metadata tags

Provenance

Source

Arxiv 250722953

Collection Method

fully automated framework

Freshness

Last updated January 2026

Geography

United States