Radiologists Spotted More on CT Scans With an AI Reading Alongside Them

Alibaba's DAMO Academy and the First Affiliated Hospital of Zhejiang University School of Medicine have published in Science a single artificial intelligence model that reads an abdominal CT scan for 146 imaging findings across 18 organs and other structures.

They also announced the open release on Sept. 18, 2026, with the model and its pre-trained checkpoints available on GitHub under an Apache 2.0 license, which allows commercial use.

The model is called RADAR. The paper was published on Sept. 17, 2026, and says that AI in radiology aims for expert-level diagnosis across many clinical tasks while existing systems remain limited in scope. In the reader study, 26 radiologists who worked with RADAR raised their diagnostic sensitivity, the share of real problems they caught, by about 10%.

RADAR was trained on more than 400,000 contrast-enhanced abdominal CT examinations, learning directly from the clinical reports radiologists had already written rather than from images labeled by hand. Across internal and external evaluations at multiple centers, the paper reports "high diagnostic performance and robust generalization," and concludes that generalist AI "can match human experts in general and complicated radiology tasks." The claim is a match with experts, not an advantage over them.