Physix Frontier badgePhysix Frontier
PublishChinese | EN

Physix Frontier · News Briefing Card (Hacker News · Sep 19, 2026)

Alibaba DAMO Academy Unveils RADAR Abdominal CT AI Model

KEY FACTS

  • Alibaba's DAMO Academy has introduced a vision-language model named RADAR.
  • The model is specifically designed for abdominal CT diagnostics, offering expert-level capabilities.
  • Training data comprises over 400,000 enhanced CT scan images.
  • It leverages 15 million anatomy-aware image-text pairs for unsupervised learning.
  • The model demonstrates superior performance in both routine and complex clinical tasks.

KEY DATA

400,000+Trained CT Scans
15,000,000Image-Text Pairs

PHYSIX OBSERVATION

RADAR directly constructs large-scale supervision signals from clinical reports, bypassing the most expensive bottleneck in medical AI: manual annotation. This pathway of extracting knowledge from unstructured text, if its generalizability is validated in real-world settings, will significantly lower the barrier to deploying specialized imaging models, driving radiology workflows from assisted reading toward fully automated initial screening.

Source: Hacker News report