(UroToday.com) The 2025 Society of Urologic Oncology (SUO) Annual Meeting was host to a bladder cancer poster session. Dr. Zorawar Singh presented a feasibility study evaluating whether a computer vision model could reliably identify anatomic landmarks within the male urethra and bladder and detect bladder tumors during cystoscopy.
As artificial intelligence (AI) becomes increasingly integrated into urologic workflows, computer vision–based systems provide an opportunity to automate recognition of both normal anatomy and pathologic findings. The study therefore aimed to determine whether an AI model could accurately recognize key urethral and bladder landmarks as well as bladder tumors on cystourethroscopy videos.
Cystoscopy videos were retrospectively analyzed using the Roboflow platform. Two independent reviewers manually annotated 652 images for six anatomic landmarks, bladder tumors, and transurethral resection (TURBT) sites. The dataset was divided into training (70%, n=416), validation (20%, n=92), and testing (10%, n=44) cohorts.
The final AI model achieved an overall mean average precision (mAP) of 76.5% with a recall of 71% across all object classes. Performance varied across anatomic categories. The model accurately detected the bladder neck with 100% mAP, normal bladder wall with 88% mAP, urethral sphincter with 86% mAP, and verumontanum with 60% mAP. Detection rates were lower for the prostatic urethra (50% mAP). Notably, the model identified bladder tumors with 100% mAP and demonstrated no false positives or false negatives in this category.
TURBT resection sites were identified with 91% mAP; however, this category showed the highest false-negative rate among all structures assessed.
The authors concluded that this pilot feasibility study demonstrates that an AI-driven computer vision model can reliably identify male urethral and bladder landmarks and detect bladder tumors with high accuracy during cystoscopy, supporting its potential for future clinical integration and real-time procedural assistance.
Presented by: Zorawar Singh, MD, Resident Physician, Department of Urology, Northwell Health, NY
Written by: Rashid K. Sayyid, MD, MSc, Assistant Professor, Urologic Oncologist, Department of Urology at The University of Arizona and Banner University Medical Center – Tucson, AZ, @rksayyid on X during the 2025 Society of Urologic Oncology (SUO) Annual Meeting, Phoenix, AZ, December 2nd–5th, 2025