Skip to content
Review Open access

Artificial Intelligence in Digital Pathology of Prostate Cancer: A Systematic Review of Clinical Validation and Pathologist Assistance

Nov 2026 · Cancers · 0 citations · 104 references

Abstract

Background/Objectives: Artificial intelligence (AI) may support prostate cancer diagnosis, but its clinical value depends on validation conditions. We synthesized evidence on its clinical validation in digital pathology and evaluated performance in cancer detection, grading, and pathologist assistance. Methods: PubMed/MEDLINE, Scopus, Web of Science Core Collection, Embase, and IEEE Xplore were searched through 28 September 2026, inclusive. The clinical synthesis included studies of digitized hematoxylin and eosin-stained histological specimens reporting cancer detection, Gleason/ISUP grading, or pathologist-assistance outcomes at the biopsy, slide, case, or patient level. Studies limited to patches were assigned to the broader evidence map. Two reviewers independently selected studies; risk of bias was assessed using QUADAS-2. Data were synthesized descriptively, without meta-analysis. Results: The broader evidence map included 562 publications; the clinical synthesis included 52 publications and 79 contributions: 34 on cancer detection, 31 on grading, and 14 on pathologist assistance or implementation. In one independent test involving 535 biopsies, the area under the ROC curve was 0.990 (95% CI 0.982–0.996). In another series, exact Grade Group agreement was recorded in 57 of 109 cancer specimens (52.3%), with a quadratic weighted kappa of 0.87. In one study, pathologists’ sensitivity increased from 88.7% to 96.6% with AI assistance. Overall risk of bias was high in 33 publications and unclear in 19. Conclusions: The findings support further clinical evaluation of AI for cancer detection, grading, and second review. Implementation requires local validation of operating thresholds and assessment of diagnostic errors and workflow impact.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.