- Published on
Alibaba open-sources AI model that can detect cancer and nearly 150 conditions
- Authors
- Name
- DeepHorus
Alibaba Group Holding’s research arm, Damo Academy, has open-sourced an artificial intelligence model named Damo Radar that can identify nearly 150 abdominal conditions, including cancers, by analyzing computed tomography (CT) scans. The announcement on Friday marks another step in Alibaba’s expanding medical AI efforts. The vision-language model is designed to examine contrast-enhanced CT scans covering 18 abdominal organs.
The model detects malignant tumors and other abnormalities. It was trained using CT scans paired with clinical reports. In nearly 40,000 real-world examinations, it achieved an average area under the curve (AUC) of 0.913 across 146 clinical findings. An AUC of 1.0 indicates perfect diagnostic accuracy.
According to a new study published in Science, the model outperformed most radiologists. This suggests AI can serve as a reliable support tool for radiologists. The open-source release allows researchers and hospitals worldwide to use and build upon the technology.
The development is significant because abdominal CT interpretation is complex and time-consuming. Radiologists face heavy workloads, and AI assistance could help prioritize urgent cases. Damo Radar’s broad coverage of 18 organs and nearly 150 conditions makes it a generalist model rather than a narrow single-disease tool.
The research team described Damo Radar as “the world’s first expert-level generalist medical imaging model.” They believe the training method could eventually be extended to other types of medical imaging. This could lead to AI systems capable of analyzing chest X-rays, MRIs, and other scans.
However, real-world clinical adoption requires further validation and regulatory approvals. Open sourcing enables independent testing and transparency. Alibaba’s move reflects a broader trend of tech companies contributing AI models to the medical community.
Looking ahead, Damo Academy plans to continue refining the model. Expanding its capabilities to more imaging modalities could improve diagnostic efficiency globally. The study’s findings offer a promising glimpse into the future of AI-assisted healthcare.
