PANORAMA study published in The Lancet Oncology

The international PANORAMA study demonstrates that high-performing artificial intelligence (AI) can improve the detection of pancreatic cancer on CT scans compared to the average radiologist. Led by Radboud University Medical Center (Radboudumc), and now published in the Lancet Oncology, it is one of the largest reader studies every performed, involving 68 radiologists from 40 institutions and 12 countries.

The researchers created a robust benchmark using a confidential dataset of nearly 400 CT scans, evaluated by a large group of international experts. Developers worldwide submitted over 250 AI models, which were systematically compared against radiologist performance.

Read the full publication here, or the news item on Radboudumc’s website.

𝗞𝗲𝘆 𝗢𝘂𝘁𝗰𝗼𝗺𝗲𝘀 The leading AI systems outperformed the average radiologist, achieving 92% accuracy compared to 88% for radiologists. The best-performing AI also generated 38% fewer false positives. Early results suggest AI could support radiologists in achieving faster and more reliable diagnoses, though further validation is required before clinical implementation.

Researchers are now developing new models using broader abdominal imaging datasets, supported by the Hanarth Fund.

Congratulations to Natália Alves, Megan Schuurmans, Anindo Saha, Pierpaolo Vendittelli, Geert Litjens, John J. Hermans, Henkjan Huisman, and the entire PANORAMA consortium for their outstanding work!