Artificial intelligence for chordee measurement in hypospadias surgery: current evidence and a blinded expert-panel validation framework.

Hypospadias-associated ventral penile curvature, commonly referred to as chordee, is a key intraoperative finding that can influence the choice of straightening technique and the need for urethral plate preservation or transection. In routine surgery, curvature is usually estimated visually during artificial erection testing, but this approach is subjective and may vary according to surgeon experience, viewing angle, erection quality, penile rotation, and image perspective. Artificial intelligence and computer vision methods may provide a more objective and reproducible way to measure curvature from standardized operative images. Early studies have used object detection, segmentation, landmark detection, Hough-transform methods, circle fitting, and geometric angle calculation. Reported technical performance is promising, particularly in controlled model-based datasets, but the clinical evidence remains limited because real intraoperative images are heterogeneous and because reference standards vary across studies. This Review summarizes the current evidence for artificial intelligence-based penile curvature measurement in hypospadias surgery and proposes a practical validation framework for future clinical studies. The most important next step is not simply improving model accuracy, but testing locked models on standardized real intraoperative images using patient-level data separation, blinded expert-panel measurements, threshold-focused analysis around 30 degrees, and direct comparison with the operating surgeon's live estimate. At present, artificial intelligence should be considered a measurement aid for documentation, quality improvement, and decision support, not a replacement for expert surgical judgment.