Alibaba Open-Sources 'Damo Radar', a Medical AI for Detecting Cancer and Over 150 Diseases

Alibaba DAMO Academy has open-sourced 'Damo Radar', a vision-language model for medical image analysis. This model analyzes contrast-enhanced CT scans to identify cancer (including malignant tumors) in 18 abdominal organs, as well as over 150 diseases and abnormal findings.

Performance and Validation

The research team evaluated the model based on approximately 40,000 real-world examination records, achieving an average AUC of 0.913 across 146 clinical findings. In a comparative study involving 26 radiologists from multiple hospitals, Damo Radar's average accuracy was higher than that of 23 specialists.

Additionally, the model's assistive effect was confirmed. When radiologists use Damo Radar, their ability to prevent missed diagnoses improves by 10%, and the time required is reduced by more than 30%.

Research Background and Scalability

The study was published in the journal Science, with participation from affiliated hospitals of Zhejiang University. The research team defines Damo Radar as the 'world's first expert-level general medical imaging model' and suggests the potential for expansion to other types of medical imaging. This achievement is part of Alibaba's medical AI strategy, following the development of pancreatic, stomach, and colorectal cancer screening tools and the 'Coca AI model' specialized for early colorectal cancer detection.