Clinical applications of artificial intelligence in genitourinary pathology

Artificial intelligence (AI) has emerged as a major area of interest in pathology, driven by improvements in whole slide imaging and digital pathology systems. This narrative review provides an overview of current and emerging AI applications in GU cancer pathology, with a focus on prostate, bladder, and renal cancers. As relevant background, we also describe the evolution of computational pathology models, from supervised and convolutional neural network-based approaches to self-supervised learning and pathology foundation models with improved generalizability. Prostate cancer represents the most mature area of development, with FDA-approved systems such as Paige Prostate, IBEX (Galen) Prostate, and the ArteraAI Prostate Test demonstrating utility in cancer detection, grading, and prognostication. In bladder cancer, AI has shown promise especially in urine cytology, where computational approaches may reduce interobserver variability and improve detection of high-grade urothelial carcinoma. In renal cell carcinoma and other less common genitourinary cancers, AI methods have demonstrated high accuracy in tumor subtype classification and grading, although clinical translation remains limited. Beyond diagnostic performance, this review also addresses barriers to clinical implementation, including regulatory pathways, diagnostic liability, economic considerations, and the need for post-deployment performance monitoring in real-world clinical settings. Overall, through prospective outcomes studies and multi-institutional validation to guide its responsible integration into clinical practice, AI has the potential to significantly improve diagnostic accuracy in GU pathology.