Abstract
BACKGROUND: This study was conducted to investigate the clinical value of a navigation system based on deep learning and mixed reality for the treatment of kidney stones with percutaneous nephrolithotomy (PNL), and to improve its theoretical basis for the treatment of kidney stones. METHODS: The data of 136 patients with kidney stones from October 2021 to December 2023 were retrospectively analyzed. All patients underwent PNL, and were categorized into a control group (Group 1) and a surgical navigation group (Group 2) according to puncture positioning method. Preoperative computed tomography (CT) was performed in both groups. In group 1, procedures were performed under standard ultrasound guidance. PNL was performed with navigation system fused with ultrasound to guide percutaneous puncture in group 2. The baseline information and procedural characteristics of both groups were compared. RESULTS: PNL was successfully performed in both groups. No significant difference was found in the baseline date between the two groups. In group 2, real-time ultrasound images could be accurately matched with CT images with the aid of navigation system. The success rate of single puncture, puncture time, and decrease in hemoglobin were significantly improved in group 2 compared to group 1. (p < 0.05). CONCLUSIONS: The application of navigation system based on deep learning and mixed reality in PNL for kidney stones allows for real-time intraoperative navigation, with acceptable accuracy and safety. Most importantly, this technique is easily mastered, particularly by novice surgeons in the field of PNL.