Release2026-09-18
Alibaba’s Damo Academy open-sourced Damo Radar. It is a vision-language model for CT scans. It covers 18 abdominal organs. It detects nearly 150 conditions including cancers. It was trained on CT scans paired with clinical reports. It achieved an average AUC of 0.913 across 146 findings in 40,000 exams. It outperformed most radiologists in testing. The team calls it the first expert-level generalist medical imaging model. The method may extend to other imaging types.
What changed
Medical imaging analysis required manual review by radiologists.
What it unlocks
Automated detection of nearly 150 abdominal conditions from CT scans.
- 146 clinical findings
- 0.913 average AUC
- 40,000 real-world exams
- 18 abdominal organs
Sources