Application of Improved Visual SLAM in Intraoperative Navigation of Robotic-Assisted Laparoscopic Surgery

In robot-assisted laparoscopic surgery, precise intraoperative navigation is essential to enhance the accuracy, safety, and efficiency of the procedure. By extracting feature points from the environment, Visual Simultaneous Localization and Mapping (V-SLAM) technology has been widely employed to accomplish self-localization and map generation. In laparoscopic surgery, V-SLAM offers high-precision real-time positional data for surgical robots, thereby improving the accuracy of operative navigation. Nonetheless, the performance of conventional V-SLAM algorithms is significantly compromised by the extremely dynamic characteristics of surgical environments. Widely used visual SLAM systems, such as ORB-SLAM, are susceptible to inaccurate feature point tracking in the presence of dynamic objects and variations in lighting, hence compromising navigation precision and surgical safety. This study assessed the efficacy of visual V-SLAM technology in enhancing intraoperative navigation during robot-assisted laparoscopic surgery, pointing out its contribution to surgical precision, safety, and operational efficiency. The research enhanced the ORB-SLAM2 algorithm to adapt to the dynamically changing conditions encountered during surgical procedures. The experiment validated the efficacy of the enhanced method within a simulated laparoscopic surgery setting and assessed the impact of three variables: light intensity, instrument occlusion, and dynamic tissue deformation, on trajectory error (ATE and RPE) and mapping accuracy. The findings indicated that under situations of low illumination, significant occlusion, and substantial deformation, ATE and RPE mistakes escalated by over 50%, while map accuracy diminished by approximately 40%, leading to heightened potential surgical risks.