RSNA 2026: AI-powered lung nodule detection achieves 97% sensitivity in multi-center trial

A landmark multi-center study presented at the Radiological Society of North America (RSNA) 2026 Annual Meeting has demonstrated breakthrough results in AI-assisted lung cancer screening.

The study, conducted across 14 hospitals in India, the United States, and Europe, evaluated an AI system trained on over 2 million chest CT examinations. The results showed a sensitivity of 97.2% for detecting clinically significant lung nodules, while simultaneously reducing average radiologist interpretation time from 8.4 minutes to 5.5 minutes per study.

Dr. Ananya Sharma from AIIMS Delhi, the lead investigator, noted that the system was particularly effective at identifying sub-solid nodules — a category where human readers typically show lower inter-observer agreement.

"This is not about replacing radiologists," Dr. Sharma emphasized. "The AI catches what we might miss on a busy day, and we catch what the AI misses due to atypical presentations. Together, the detection rate exceeds either alone."

The study is expected to influence upcoming Indian government guidelines on CT lung screening programs, which are currently being developed for high-risk populations in industrial cities.