Alibaba open-sources RADAR AI model that detects cancer from CT scans

Alibaba DAMO Academy and Zhejiang University's First Affiliated Hospital published in Science and open-sourced under Apache 2.0 the code for the DAMO RADAR vision-language model, which detects 146 clinical findings across 18 organs from contrast-enhanced abdominal CT scans with a mean AUC of 0.913.

Compiled by Sergey KostenchukPublished 2026-09-18Updated 2026-09-19

2026-09-18 Research HuggingFace

🤖 Alibaba open-sources RADAR AI model that detects cancer from CT scans

Alibaba DAMO Academy and Zhejiang University's First Affiliated Hospital published the code for the DAMO RADAR model in Science — an AI for analyzing abdominal CT scans. The model was trained on 400,000 CT scans (15 million 'anatomy-text' pairs) from clinical reports, without manual annotation, and covers 146 findings across 18 organs. Mean AUC is 0.913; in a blind comparison, it was more accurate than 23 of 26 radiologists.

🌍 Claimed first 'expert' generalist approach: open code under Apache 2.0 allows hospitals and startups to fine-tune and validate RADAR.

👤 Code and checkpoints are already on GitHub and Hugging Face, results can be reproduced on demo data; this is a research model, not a certified medical product.

Source 1: https://www.scmp.com/tech/big-tech/article/3368055/alibaba-open-sources-medical-ai-model-can-detect-cancer-and-nearly-150-conditions Source 2: https://www.science.org/doi/10.1126/science.aec6129

Sources