Alibaba DAMO Academy Releases AI Model DAMO EAGLE for Esophageal Cancer Screening

Alibaba DAMO Academy, together with Sichuan Cancer Hospital and Sun Yat-sen University Cancer Center, has developed an AI model called DAMO EAGLE for esophageal cancer screening. The model can identify esophageal cancer and pre-cancerous lesions from plain chest CT scans without endoscopy or contrast agents. It has been validated on over 80,000 cases across 3 countries, with the related paper published in Nature Medicine on September 22.

Esophageal cancer in China accounts for about half of global cases and deaths, with most patients diagnosed at late stages. Standard screening via endoscopy is invasive and complex, making it hard to scale. DAMO EAGLE overcomes the challenge by pairing lesion locations from endoscopy reports and contrast CT scans onto plain CT images to train the AI.

In opportunistic screening, the model achieved 90% sensitivity for esophageal cancer, 52.5% for subtle pre-cancerous lesions, and 99.2% specificity. Performance held up on low-dose CT, allowing seamless integration into existing lung cancer screening. The team says AI is not meant to replace endoscopy but to filter high-risk individuals for more precise testing.