Document is current

Article History

Received: 8 June 2026

Revised: 30 August 2026

Accepted: 1 September 2026

First Online: 26 September 2026

Compliance with ethical standards

Guarantor: The scientific guarantor of this publication is Karsten Guenzel, MD.

Conflict of interest: The authors of this manuscript declare relationships with the following companies: M.G.P. and A.v.L. are employees of DeepHealth Inc., the manufacturer of the evaluated CADe/x device. F.G. reports consulting fees from DeepHealth, SpectraCure, and Procept outside of the submitted work, and speaker fees from Bayer and Siemens. He is not an employee of DeepHealth and was not involved in the conceptualisation or development of the commercial product evaluated in this study. V.P. and F.G. are members of the Scientific Editorial Board of European Radiology (section: Urogenital) and, as such, did not participate in the selection or review processes for this article. K.G., E.M., M.P. and R.P. declare no conflicts of interest related to this work.

Statistics and biometry: One of the authors has significant statistical expertise.

Informed consent: At the participating institutions, informed consent was either waived by the local IRB or obtained at the time of MRI examination through institutional procedures allowing future use of clinical data for research purposes.

Ethical approval: Institutional Review Board approval was obtained.

Study subjects or cohorts overlap: Portions of the Vivantes and St. Antonius cohorts have been described previously: Van den Berg et al [ ], Guenzel et al [ ]. The present study addresses a different question by evaluating commercial AI risk classification for csPCa detection in a pooled multicentre, multivendor cohort, including patient-level non-inferiority and ROC analyses, lesion-level FROC analysis, and modeled biopsy-triage strategies. None of the imaging or pathology data in the present study were used to develop or train the evaluated AI system.

Methodology: