Artificial Intelligence in Functional Urology: How It may Shape the Future

Imad Bentellis; Sonia Guérin; Zine-Eddine Khene; Rose Khavari; Benoit Peyronnet


Curr Opin Urol. 2021;31(4):385-390. 

In This Article

Surgical Training/Surgical Performance

The use of robotic surgery in functional urology has expanded significantly over the past decade.[23] Its computer interface provides privileged access to record automated performance metrics (APMs).[24] APMs related to instrument motions (path length, instrument speed, and articulations), camera motions, and actions such as energy use. Combined APMs and clinical features, Hung and al could predict continence (0 or 1 pads) after robot-assisted prostatectomy (RAPN) with an accuracy of 0.6. This refined analysis of instrument motions may be used to improve surgeon's training.

In addition, AI methods have been used to recognize surgical steps from video/endoscopic procedures. Zia et al. could identify RAPN steps with an average precision of 80.9% with a convolutional neuro network.[25] Beyond the surgery steps, Khalid et al. showed suturing, knot tying, and needle passing could be automatically identified using the JHU-ISI Gesture and Skill Assessment Working Set (JIGSAWS) with an accuracy higher than 0.99.[26] In the future, we speculate that APMs combined with automatic recognition could be an effective tool for surgeons' technical skills evaluation. Nevertheless, other studies are required to validate that approach.