External validation of an artiﬁcial intelligence model for Gleason grading of prostate cancer on prostatectomy specimens

Sep 21, 2026

External validation evaluates the DeepDx Prostate AI algorithm for Gleason grading on whole-mount prostate histopathology, addressing domain shift between biopsy-trained models and radical prostatectomy (RP) specimens with different tissue representation and sample sizes. The study applies the AI system to tiles derived from RP specimens and compares predicted Gleason patterns and grade groups against expert uropathologist reference grading. Agreement is quantified using Cohen’s kappa and performance metrics for cancer identification and grade grouping.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

Objectives

Materials and Methods

Results

Conclusion

Introduction

BJU Int 2025; 135: 133–139 doi:10.1111/bju.16464 Original Article External validation of an artiﬁcial intelligence model for Gleason grading of prostate cancer on prostatectomy specimens Bogdana Schmidt 1, Simon John Christoph Soerensen2,3 , Hriday P. Bhambhvani7, Richard E. Fan2, Indrani Bhattacharya4, Moon Hyung Choi9 , Christian A. Kunder5, Chia-Sui Kao8, John Higgins5, Mirabela Rusu2,4,6 and Geoffrey A. Sonn2,4 1Division of Urology, Department of Surgery, Huntsman Cancer Hospital, University of Utah, Salt Lake City, UT, 2Department of Urology, 3Department of Epidemiology and Population Health, 4Department of Radiology, 5Department of Pathology, Stanford University School of Medicine, 6Department of Biomedical Data Science, Stanford University, Stanford, CA, 7Department of Urology, Weill Cornell Medical College, New York-Presbyterian Hospital, New York, NY, 8Department of Pathology and Laboratory Medicine, Cleveland Clinic, Cleveland, OH, USA, and 9Department of Radiology, College of Medicine, Eunpyeong St. Mary’s Hospital, The Catholic University of Korea, Seoul, Korea B.S. and S.J.C.S. shared ﬁrst authorship. M. R. and G.A.S. shared senior authorship. Objectives To externally validate the performance of the DeepDx Prostate artiﬁcial intelligence (AI) algorithm (Deep Bio Inc., Seoul, South Korea) for Gleason grading on whole-mount prostate histopathology, considering potential variations observed when applying AI models trained on biopsy samples to radical prostatectomy (RP) specimens due to inherent differences in tissue representation and sample size. Materials and Methods The commercially available DeepDx Prostate AI algorithm is an automated Gleason grading system that was previously trained using 1133 prostate core biopsy images and validated on 700 biopsy images from two institutions. We assessed the AI algorithm’s performance, which outputs Gleason patterns (3, 4, or 5), on 500 1-mm2 tiles created from 150 whole-mount RP specimens from a third institution. These patterns were then grouped into grade groups (GGs) for comparison with expert pathologist assessments. The reference standard was the International Society of Urological Pathology GG as established by two experienced uropathologists with a third expert to adjudicate discordant cases. We deﬁned the main metric as the agreement with the reference standard, using Cohen’s kappa. Results The agreement between the two experienced pathologists in determining GGs at the tile level had a quadratically weighted Cohen’s kappa of 0.94 . The agreement between the AI algorithm and the reference standard in differentiating cancerous vs non-cancerous tissue had an unweighted Cohen’s kappa of 0.91 . Additionally, the AI algorithm’s agreement with the reference standard in classifying tiles into GGs had a quadratically weighted Cohen’s kappa of 0.89 . In distinguishing cancerous vs non-cancerous tissue, the AI algorithm achieved a sensitivity of 0.997 and speciﬁcity of 0.88; in classifying GG ≥2 vs GG 1 and non-cancerous tissue, it demonstrated a sensitivity of 0.98 and speciﬁcity of 0.85 . Conclusion The DeepDx Prostate AI algorithm had excellent agreement with expert uropathologists and performance in cancer identiﬁcation and grading on RP specimens, despite being trained on biopsy specimens from an entirely different patient population. Keywords radical prostatectomy, artiﬁcial intelligence, Gleason grading, prostate cancer, whole-mount prostatectomy 􀀁 2024 The Author(s) . BJU International published by John Wiley & Sons Ltd on behalf of BJU [International. www.bjui.org](International. www.bjui.org) [wileyonlinelibrary.com](wileyonlinelibrary.com) This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modiﬁcations or adaptations are made. Introdu

FAQs

What was the study’s primary goal for the DeepDx Prostate AI model?

To externally validate DeepDx Prostate for Gleason grading on whole-mount prostate histopathology, accounting for differences between biopsy-trained models and radical prostatectomy specimens.

How was the AI evaluated in the materials and methods?

DeepDx Prostate was assessed on 1-mm² tiles created from whole-mount RP specimens, with outputs converted into grade groups for comparison to expert pathologist assessments using Cohen’s kappa.

What were the key results for agreement and cancer identification?

The AI showed strong agreement with the reference standard, including high Cohen’s kappa values for grade group classification and strong sensitivity for cancer vs non-cancer detection, with reported specificity values as well.

Sep 21, 2026

Uploaded by Evangeline