Artificial intelligence and radiologists in pancreatic cancer detection using standard of care CT scans (PANORAMA): an international, paired, non-inferiority, confirmatory, observational study - 23/12/25

Doi : 10.1016/S1470-2045(25)00567-4

Natalia Alves, MSc ^a,^{*
Joint first authors}⁎ , Megan Schuurmans, MSc ^a,^{*
Joint first authors}, Dawid Rutkowski, MSc ^c, Anindo Saha, MSc ^a, Pierpaolo Vendittelli, MSc ^a, Nancy Obuchowski, ProfPhD ^d, Marjolein H Liedenbaum, MD ^e, Ingfrid S Haldorsen, ProfMD ^e, Anders Molven, ProfPhD ^f, Derya Yakar, ProfMD ^g, Jeroen Geerdink, BSc ^h, Sebastiaan van Koeverden, MD ⁱ, Deniece M Riviere, MD ⁱ, Wulphert Venderink, MD ⁱ, Robbert de Haas, MD ^g, Namkug Kim, ProfPhD ^j, J-Matthias Löhr, ProfMD ^c, Garima Suman, MD ^k, Klaus H Maier-Hein, ProfPhD ^l, Horst K Hahn, ProfPhD ^m, Weichung Wang, ProfPhD ⁿ, Alan L Yuille, ProfPhD ^o, Avinash Kambadakone, MD ^p, Elliot K Fishman, ProfMD ^q, Caroline Verbeke, ProfMD ^r, Geert Litjens, ProfPhD ^a,^b, John J Hermans, MD ^i,^{^{†
Contributed equally}}, Henkjan Huisman, ProfPhD ^a,^{^{†
Contributed equally}}
on behalf of the
PANORAMA consortium ^{^‡}
For the complete list of the PANORAMA consortium members, see the Supplementary Material )

Natália Alves, Megan Schuurmans, Anindo Saha, Pierpaolo Vendittelli, Geert Litjens, John Hermans, Henkjan Huisman, Deniece M. Riviere, Wulphert Venderink, Sebastiaan van Koeverden, Dawid Rutkowski, Marjolein H. Liedenbaum, Ingfrid S. Haldorsen, Anders Molven, Derya Yakar, Robbert J. de Haas, Jeroen Geerdink, Jeroen Veltman, Alan Yuille, Avinash Kambadakone, Caroline Verbeke, Celso Matos, Elliot Fishman, Garima Suman, Horst K. Hahn, Klaus Maier-Hein, J-Matthias Löhr, Namkug Kim, Nancy Obuchowski, Steven Gallinger, Weichung Wang, Ali Stunt, Han Liu, Riqiang Gao, Sasa Grbic, Zengtian Deng, Yimeng He, Yu Shi, Rebeca Vétil, Noëlie Debs, Clément Abi-Nader, Alexandre Bône, Marc-Michel Rohé, Ching-Yuan Yu, Jun Ma, Tianhao Fu, Bo Wang, Abraham Fourie Bezuidenhout, Adrian Thomas Huber, Adriano Liguori, Amine Korchi, Andrea Ponsiglione, Anselm Schulz, Arnaldo Stanzione, Augusto Minieri, Bang-Bin Chen, Cesare Maino, Charikleia Triantopoulou, Dimitra Christodoulou, Dominik Geisel, Dow-Mu Koh, Elisa Boffa, Enrico Boninsegna, Enza Genco, Erik Soloff, Eugenia Amelia Lettieri, Federica Omboni, Francesca Castagnoli, Francesco Prato, Frank Wessels, Giacomo Avesani, Giorgia Porrello, Giorgio Brembilla, Giovanni Morana, Giulia Zamboni, Giuseppe di Costanzo, Gunnar Juliusson, Håvard Bjørke Jenssen, Herman Zandvoort, Jeroen Pijls, Jip Prince, Katja De Paepe, Kosta Petrovic, Loekie van Valkenhoef, Luca Fortuna, Luigi Mannacio, Marc Engelbrecht, Marco Chincarini, Marco Dioguardi Burgio, Marta Zerunian, Massimo Imbriaco, Matilde Bariani, Matteo Bonatti, Maxime Ronot, Natalie Norstedt, Nazmi Kurt, Nirav Patel, Paolo Maria Sbeghen, Pawan Patel, Pietro Andrea Bonaffini, Raffaella Pozzi Mucelli, Raşit Eren Büyüktoka, Remy Geenen, Renato Cuocolo, Riccardo Valletta, Roberta Musella, Roberto Cannella, Roy S. Dwarkasing, Silvia Venturini, Sofia Gourtsoyianni, Sonaz Malekzadeh, Umberto Tupputi, Verena Obmann, Vivi Liu

^a Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen, The Netherlands

^b Oncode Institute, Utrecht, the Netherlands

^c Department of Surgery and Oncology, Karolinska University Hospital, Stockholm, Sweden

^d Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, OH, USA

^e Department of Radiology, Haukeland University Hospital, Bergen, Norway

^f Department of Clinical Medicine, University of Bergen, Bergen, Norway

^g Department of Radiology, University Medical Center Groningen, Groningen, The Netherlands

^h Ziekenhuisgroep Twente, Almelo, The Netherlands

ⁱ Department of Medical Imaging, Radboud University Medical Center, Nijmegen, The Netherlands

^j Department of Convergence Medicine, University of Ulsan College of Medicine, Asan Medical Center, Seoul, South Korea

^k Department of Radiology, Mayo Clinic, Rochester, MN, USA

^l Division of Medical Image Computing, German Cancer Research Center Heidelberg, Heidelberg, Germany

^m University of Bremen and Fraunhofer Institute for Medical Imaging MEVIS, Bremen, Germany

ⁿ Institute of Applied Mathematical Sciences, National Taiwan University, Taipei, Taiwan

^o Department of Computer Science, Johns Hopkins University, Baltimore, MD, USA

^p Department of Radiology, Massachusetts General Hospital, Boston, MA, USA

^q Department of Radiology, Johns Hopkins Hospital, Baltimore, MD, USA

^r Department of Pathology, Institute of Clinical Medicine, University of Oslo, Oslo, Norway

^* Correspondence to: Natalia Alves, Diagnostic Image Analysis Group, Radboud University Medical Center, Nijmegen 6525 GA, The Netherlands Diagnostic Image Analysis Group Radboud University Medical Center Nijmegen GA 6525 The Netherlands

Summary

Background

Pancreatic ductal adenocarcinoma (PDAC) has the worst prognosis among major cancer types, primarily due to late diagnosis on contrast-enhanced CT. Artificial intelligence (AI) can improve diagnostic performance, but robust benchmarks and reliable comparison to radiologists’ performance are scarce. We established an open-source benchmark with the aim of investigating AI systems for PDAC detection on CT and compared them to radiologists’ performance, at scale.

Methods

In this international, paired, non-inferiority, confirmatory, observational study (PANORAMA), the AI system was trained and externally validated within an international benchmark, with a cohort of 2310 patients from four tertiary care centres in the Netherlands and the USA for training (n=2224) and tuning (n=86), and a sequestered cohort of 1130 patients from five tertiary care centres (the Netherlands, Sweden, and Norway) for testing. A multi-reader, multi-case observer study with 68 radiologists (40 centres, 12 countries; median 9·0 [IQR 6·0–14·5] years of experience) was conducted on a subset of 391 patients from the testing cohort. The reference standard was established with histopathology and at least 3 years of clinical follow-up. The primary endpoint was the mean area under the receiver operating characteristic curve (AUROC) of the AI system compared to that of radiologists at PDAC detection on CT. The study protocol and statistical plan were prespecified to test non-inferiority (considering a margin of 0·05), followed by superiority towards the AI system. This study is registered with Zenodo ( zenodo.10599559 ) and is complete.

Findings

Of the 3440 (1511 [44%] female, 1929 [56%] male; median age 67 [IQR 58–74] years) included patients (Jan 1, 2004 to Dec 31, 2023), 1103 (32%) received a positive PDAC diagnosis. In the sequestered testing cohort of 1130 patients (406 with histologically confirmed PDAC), AI achieved an AUROC of 0·92 (95% CI 0·90–0·93). In the subset of 391 patients (144 [37%] with histologically confirmed PDAC) used for the reader study, AI achieved statistically non-inferior (p<0·0001) and superior (p=0·001) performance with an AUROC of 0·92 (95% CI 0·89–0·94), compared to the pool of 68 participating radiologists, with an AUROC of 0·88 (0·85–0·91).

Interpretation

AI demonstrated substantially improved PDAC detection on routine CT scans compared to radiologists on average, showing potential to detect cancer earlier and improve patient outcomes.

Funding

European Union’s Horizon 2020 research and innovation programme.

Le texte complet de cet article est disponible en PDF.

Plan

Introduction

Methods

Study design and patient cohorts

Role of the funding source

Export

Vol 27 - N° 1

P. 116-124 - janvier 2026 Retour au numéro

Article précédent

Tiragolumab plus atezolizumab and chemotherapy as first-line treatment for patients with unresectable oesophageal squamous cell carcinoma (SKYSCRAPER-08): a randomised, double-blind, placebo-controlled, phase 3 trial
Chih-Hung Hsu, Zhihao Lu, Shegan Gao, Junye Wang, Jong-Mu Sun, Tianshu Liu, Qingxia Fan, Jun Cai, Feijiao Ge, Sijing Li, Li Zhang, Edward Cha, Simon Allen, Lin Shen

| Article suivant

Electronic health record-facilitated symptom surveillance and collaborative care intervention in oncology (E2C2): a cluster-randomised, population-level, stepped-wedge, pragmatic trial
Andrea L Cheville, Jeph Herrin, Deirdre R Pachman, Veronica Grzegorczyk, Kurt Kroenke, Jennifer L Ridgeway, Sarah A Minteer, Jessica D Austin, Joan M Griffin, Linda Chlan, Cindy Tofthagen, Sandra A Mitchell, Ashley Smith, Kathryn J Ruddy

Bienvenue sur EM-consulte, la référence des professionnels de santé.
L’accès au texte intégral de cet article nécessite un abonnement.

Déjà abonné à cette revue ?

connectez-vous ou créez un compte

Artificial intelligence and radiologists in pancreatic cancer detection using standard of care CT scans (PANORAMA): an international, paired, non-inferiority, confirmatory, observational study - 23/12/25

Summary

Background

Methods

Findings

Interpretation

Funding

Plan

Export citations

Fichier

Contenu

Accès rapides

Mon compte

Aide & support

Plateformes Elsevier Masson

Déclaration CNIL