Skip to main content

Alibaba's AI Spots Esophageal Cancer on Routine CT Scans

Alibaba's AI Spots Esophageal Cancer on Routine CT Scans

Alibaba DAMO Academy has just released its fourth cancer-screening AI model, and this one tackles a notoriously tricky cancer. Called DAMO EAGLE, the model can identify esophageal cancer—including early-stage and precancerous lesions—from plain CT scans. No endoscopy, no contrast agents. Just a regular chest CT.

Developed with Sichuan Cancer Hospital and Sun Yat-sen University Cancer Center, the model was validated on 80,000 cases across three countries. The findings appeared in Nature Medicine on September 22.

Image

Cracking the 'Impossible' Task

Why does this matter? China accounts for nearly half of the world's esophageal cancer cases and deaths. Most patients are diagnosed late, missing the window for curative treatment. But if caught early, the five-year survival rate can exceed 95%.

The standard screening tool is upper gastrointestinal endoscopy—effective but invasive, uncomfortable, and hard to scale. Plain chest CT, on the other hand, is cheap, well-accepted, and already used for lung cancer screening. Its scan range naturally covers the esophagus. So why not use it to spot esophageal cancer too?

The challenge: the esophagus is a long, collapsible tube. Early lesions hide in the mucosal lining, and the heart's constant beating blurs the picture. Radiologists struggle to see anything suspicious on plain CT.

Yao Jiawen, an algorithm expert at DAMO Academy, explained how the team solved it. They matched endoscopic reports with precise lesion locations on enhanced CT, then mapped those onto plain CT images. This trained the AI to pick up early signs the human eye would miss.

The results are striking. In opportunistic screening, DAMO EAGLE achieved 90% sensitivity for esophageal cancer—meaning it rarely misses a case. For precancerous signals, which are extremely faint, sensitivity reached 52.5%. In real-world settings, specificity hit 99.2%, so false positives are rare. And it performs just as well on low-dose CT, making it easy to slot into existing lung cancer screening programs without extra burden on patients.

Completing the 'Plain CT + AI' Route

Wang Qifeng, chief physician at Sichuan Cancer Hospital's Radiotherapy Department, noted that western China—especially Sichuan—has a high incidence of esophageal cancer, partly due to diets heavy in hot, salty foods. But endoscopic screening is limited by resources and patient compliance. "AI isn't replacing endoscopy," Wang said. "It's identifying high-risk individuals who then get endoscopy, making the exam more targeted and boosting detection rates." More importantly, it pushes prevention earlier, opening more doors for early treatment.

With models for pancreatic cancer, gastric cancer, colorectal cancer, esophageal cancer, and aortic dissection already released, DAMO Academy has now built out its "plain CT + AI" multi-check platform. Five papers in Nature Medicine later, the vision is clear: a single plain CT scan could one day screen for China's top seven deadliest cancers.

Key Points:

  • DAMO EAGLE detects esophageal cancer from plain CT scans without endoscopy or contrast agents.
  • Validated on 80,000 cases across three countries; published in Nature Medicine on September 22.
  • Achieves 90% sensitivity for esophageal cancer and 99.2% specificity in real-world settings.
  • Works on low-dose CT, allowing seamless integration into lung cancer screening.
  • Part of Alibaba's broader "plain CT + AI" strategy to screen for multiple cancers at once.