When matched for sensitivity, the AI system produced fewer false positives than radiologists. Experts suggest that AI may enable earlier and more accurate diagnosis of pancreatic cancer. Overall, the AI system outperformed radiologists in identifying pancreatic ductal adenocarcinoma (PDAC) on routine CT scans in this observational study.
The findings are from the PANORAMA study, an international paired non-inferiority trial published in The Lancet Oncology. Led by Natalia Alves and colleagues from the Diagnostic Image Analysis Group at Radboud University Medical Center, the investigation was designed to create a robust benchmark for PDAC detection using contrast-enhanced CT scans and determine how modern AI algorithms compare with radiologists at scale. PDAC is known for having one of the poorest survival rates among major cancers, largely because diagnosis often occurs too late for curative treatment.
The study aimed to address this diagnostic challenge by evaluating whether AI could enhance detection accuracy earlier in the disease course.
To achieve this, the team developed and validated an AI model using a large, diverse dataset. A total of 2310 patients from tertiary centers in the Netherlands and the USA contributed to the training and tuning phases. An independent cohort of 1130 patients from centers in the Netherlands, Sweden, and Norway was then used for testing. In addition, 68 radiologists from 12 countries participated in a multi-reader, multi-case assessment involving 391 patient scans to allow a direct comparison between human performance and AI output. Diagnoses were confirmed using histopathology and a minimum of three years of clinical follow-up.
The key findings of the study were as follows:
- Of the 3440 patients assessed across all datasets, 1103 were diagnosed with pancreatic ductal adenocarcinoma.
- In the main testing cohort, the AI system showed strong diagnostic performance with an AUROC of 0.92.
- In the paired reader study, the AI model again achieved an AUROC of 0.92, confirming statistically significant non-inferiority and superiority compared with radiologists.
- The pooled performance of the 68 radiologists resulted in an AUROC of 0.88.
- When sensitivity was matched between groups, the AI system generated fewer false positives, demonstrating better precision in distinguishing PDAC from non-cancer cases.
These results demonstrate clear potential for integrating AI into pancreatic cancer diagnostics. By improving accuracy on standard-of-care CT scans, AI could support earlier detection—an essential factor in improving outcomes for a cancer known for its rapid progression and late presentation. The authors note that AI is not intended to replace radiologists but can serve as a powerful adjunct tool, particularly in busy clinical settings or regions facing specialist shortages.
The PANORAMA study provides one of the strongest validations to date for AI-assisted PDAC detection and offers an open-source benchmark that future research can build upon. As healthcare systems increasingly adopt AI technologies, such evidence-based evaluations will be critical for ensuring safe, effective clinical implementation.
Reference: https://www.thelancet.com/journals/lanonc/article/PIIS1470-2045(25)00567-4/abstract
Disclaimer: This website is primarily for healthcare professionals. The content here does not replace medical advice and should not be used as medical, diagnostic, endorsement, treatment, or prescription advice. Medical science evolves rapidly, and we strive to keep our information current. If you find any discrepancies, please contact us at corrections@medicaldialogues.in. Read our Correction Policy here. Nothing here should be used as a substitute for medical advice, diagnosis, or treatment. We do not endorse any healthcare advice that contradicts a physician's guidance. Use of this site is subject to our Terms of Use, Privacy Policy, and Advertisement Policy. For more details, read our Full Disclaimer here.
NOTE: Join us in combating medical misinformation. If you encounter a questionable health, medical, or medical education claim, email us at factcheck@medicaldialogues.in for evaluation.