Inclusion criteria involved the selection of articles reporting the treatment of adult patients with urolithiasis or different models of urolithiasis. In the study group, endoscopic lithotripsy (RIRS, FURS, or PCNL) with the help of AI was performed, while the same procedure without AI assistance was performed in the comparison group. The investigated results included intra- and postoperative complications, stone-free rate, operation time, and hospital stay. Exclusion criteria were non-English publications, reviews, case reports, and studies without an intraoperative AI application. As a result, 6 articles were included. Selected prospective and retrospective studies cover the period from 2018 to 2023 and use various datasets, including in vivo models on animals, in vitro phantoms, clinical videos, and image repositories. The main ways of using AI in these works can be categorized into three domains: intraoperative navigation, tissue and stone differentiation, and stone classification according to chemical composition.
A single study by Fu et al.1 focused on intraoperative navigation. In addition to standard surgical instruments, their system required the Aurora electromagnetic (EM) tracking system with an EM probe fixed at the tip of the ureteroscope. The system allowed for real-time tracking of the ureteroscope tip, which enabled reconstruction of the 3D map of endoscope movement.
A total of four studies explored the use of AI for intraoperative segmentation (differentiation) of kidney stones and urinary tract tissues during endourological procedures. Schlager et al.2 and Jeong et al.3 developed AI-assisted systems in order to avoid collateral damage during laser firing by reliably differentiating between anatomical structures, such as stones, tissues, and instruments. Gupta et al.4 and Setia et al.5 suggest segmentation of kidney stones, their fragments, and laser fibers during surgery to measure stone size, which may be useful in deciding whether further lithotripsy is required, and also for intraoperative tracking.
Lopez-Tiro et al.6 evaluated several machine learning and deep learning methods for in vivo classification of kidney stones using endoscopic images, aiming to identify stone composition in real time during ureteroscopy and reduce the need for fragment extraction and delayed analysis.
Besides the abovementioned applications that were described in the included articles, hypothetically, use of AI per-operatively can assist in: real-time navigation and guidance; automated instrument tracking; wearable sensor integration for improving surgical dexterity.
In conclusion, the systematic review revealed several promising approaches to the intraoperative application of AI. Preliminary results of the reviewed articles demonstrate high accuracy in real time, which indicates a potential for AI-based development. Their implementation may enhance patients' safety, improve surgical accuracy, and lead to better treatment outcomes.
Written by: Andrey Morozov,1 Valeria Patrikeeva,2 Igor Matkovskiy,2 Bhaskar Somani,3 Jack Baniel,4,5 Olivier Traxer,6 David Lifshitz,4,5 Stanislav Ali,1 Yaron Erlich,4,5 Shay Golan,4 Dmitry Pushkar,7 Giovanni E. Cacciamani,8,9 Vineet Gauhar,10,11 Dmitry Enikeev,1,4,5,12
- Institute for Urology and Reproductive Health, Sechenov University, Moscow, Russia
- N. V. Sklifosovskiy Institute of Clinical Medicine, Sechenov University, Moscow, Russia
- Department of Urology, University Hospital Southampton NHS Trust, Southampton, UK.
- Department of Urology, Rabin Medical Center, Petah Tikva, Israel
- Faculty of Medical and Health Sciences, Tel Aviv University, Tel Aviv, Israel
- Department of Urology, Tenon Hospital, Hôpitaux Universitaires Paris-Est, AP-HP, Université Pierre et Marie Curie Paris, Paris, France.
- Urology Department, Moscow State University of Medicine and Dentistry, Moscow, Russia
- USC Institute of Urology and Catherine & Joseph Aresty Department of Urology, Keck School of Medicine
- Artificial Intelligence Center at USC Urology, USC Institute of Urology, University of Southern California, Los Angeles, California, USA
- Ng Teng Fong General Hospital, NUH, Singapore
- Asian Institute of Nephrourology (AINU), India
- Department of Urology, Medical University of Vienna, Vienna, Austria
- Z. Fu et al., ‘Visual-electromagnetic system: A novel fusion-based monocular localization, reconstruction, and measurement for flexible ureteroscopy’, International Journal of Medical Robotics and Computer Assisted Surgery, vol. 17, no. 4, pp. 1–16, 2021, doi: 10.1002/rcs.2274.
- D. Schlager, J. Schütz, A. Brandenburg, and A. Miernik, ‘Automated stone/tissue autofluorescence analysis in real-time – an ex vivo evaluation of an intelligent laser lithotripsy system’, European Urology Supplements, vol. 17, no. 2, p. e1788, Mar. 2018, doi: 10.1016/S1569-9056(18)32236-X.
- Jeong, K. Chang, J. Lee, and J. Choi, ‘A warning system for urolithiasis via retrograde intrarenal surgery using machine learning: an experimental study’, BMC Urol, vol. 22, no. 1, p. 80, Dec. 2022, doi: 10.1186/S12894-022-01032-5.
- S. Gupta, S. Ali, L. Goldsmith, B. Turney, and J. Rittscher, ‘Multi-class motion-based semantic segmentation for ureteroscopy and laser lithotripsy’, Computerized Medical Imaging and Graphics, vol. 101, p. 102112, Oct. 2022, doi: 10.1016/J.COMPMEDIMAG.2022.102112.
- S. A. Setia, Z. A. Stoebner, C. Floyd, D. Lu, I. Oguz, and N. L. Kavoussi, ‘Computer Vision Enabled Segmentation of Kidney Stones During Ureteroscopy and Laser Lithotripsy’, J Endourol, vol. 37, no. 4, pp. 495–501, Apr. 2023, doi: 10.1089/END.2022.0511.
- F. Lopez et al., ‘Assessing deep learning methods for the identification of kidney stones in endoscopic images’, Annu Int Conf IEEE Eng Med Biol Soc, vol. 2021, pp. 2778–2781, 2021, doi: 10.1109/EMBC46164.2021.9630211.