A large international study has found that an artificial intelligence (AI) system can detect pancreatic ductal adenocarcinoma (PDAC) on CT scans more accurately than experienced radiologists. The findings could pave the way for earlier cancer detection and improved patient outcomes for one of the deadliest malignancies.
Benchmark Study Across Multiple Centres
The PANORAMA study, an international, paired, non-inferiority, confirmatory, observational project, evaluated an AI system trained to identify PDAC on contrast-enhanced CT scans. Researchers established an open-source benchmark to allow reliable comparison between AI and human experts.
The study involved 3,440 patients from nine tertiary care centres across the Netherlands, the USA, Sweden, and Norway between January 2004 and December 2023. Of these, 1,103 patients (32%) were diagnosed with PDAC based on histopathology and long-term clinical follow-up.
A cohort of 2,310 patients was used to train and validate the AI system, while a separate testing cohort of 1,130 patients ensured unbiased performance evaluation. In addition, 68 radiologists from 40 centres in 12 countries participated in a controlled observer study, interpreting CT scans from a subset of 391 patients under the same conditions.
Superior Accuracy To Human Readers
The AI system achieved a mean area under the receiver operating characteristic curve (AUROC) of 0.92 (95% CI 0.90–0.93) in the test cohort, compared with the pooled performance of radiologists who achieved an AUROC of 0.88 (0.85–0.91). The difference was statistically significant, confirming both non-inferiority (p<0.0001) and superiority (p=0.001) of the AI system.
Implications For Clinical Practice
PDAC has the poorest prognosis among major cancers, largely due to late diagnosis. The study’s authors suggest that integrating AI systems into routine radiology workflows could help identify early-stage PDAC more consistently and reduce diagnostic delays.
Reference
Alves N et al. Artificial intelligence and radiologists in pancreatic cancer detection using standard of care CT scans (PANORAMA): an international, paired, non-inferiority, confirmatory, observational study. The Lancet Oncology. 2026;27(1):116-24.
