Li Dongyuan, Shan Yixin, Shi Haochen, Tu Puxun, Huang Shenghao, Sun Weiyan, Zhao Deping, Chen Chang, Chen Xiaojun
Institute of Biomedical Manufacturing and Life Quality Engineering, State Key Laboratory of Mechanical System and Vibration, School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200241, China.
Department of Thoracic Surgery, Shanghai Pulmonary Hospital, School of Medicine, Tongji University, Shanghai 200433, China.
Med Image Anal. 2025 Oct;105:103691. doi: 10.1016/j.media.2025.103691. Epub 2025 Jun 23.
The precise localization of small peripheral lung nodules (< 2 cm) poses a frequent challenge in clinical practice owing to the minimal invasion and low radiation dose requirements. The accuracy of lesion localization is often affected by the patient's respiratory movement during the localization process. In this study, an enhanced robot-assisted localization framework that integrates a novel motion compensation algorithm is proposed for radiation-free pulmonary nodule localization. Specifically, an algorithm for preoperative semi-automatic recognition of surface feature points based on point cloud optimization and clinical knowledge is proposed. The algorithm allows the optimized feature point distribution by minimizing the registration error of different distribution types and random initializations. Then, the laser-world hand-eye calibration algorithm is proposed to achieve accurate calibration of the system. Finally, motion compensation is applied during intraoperative registration to reduce the error caused by the physiological movement of the patient. The accuracy and feasibility of the system were verified through phantom, volunteer, and clinical experiments. The results demonstrate that our proposed automatic preoperative surface feature recognition, intraoperative registration, and localization methods provide a highly potential solution for clinical lesion localization. This study advanced the existing navigation and clinical application protocols of thoracic surgery robotic systems by implementing an innovative and robust positioning method for chest lesions. The integrated system was successfully validated in a clinical setting, demonstrating its potential to transform minimally invasive thoracic surgical procedures by simultaneously minimizing radiation exposure and complications.
由于需要最小的侵入性和低辐射剂量，小的外周肺结节（<2 cm）的精确定位在临床实践中经常构成挑战。在定位过程中，病变定位的准确性常常受到患者呼吸运动的影响。在本研究中，提出了一种集成了新型运动补偿算法的增强型机器人辅助定位框架，用于无辐射的肺结节定位。具体而言，提出了一种基于点云优化和临床知识的术前表面特征点半自动识别算法。该算法通过最小化不同分布类型和随机初始化的配准误差来实现优化的特征点分布。然后，提出了激光-世界手眼校准算法以实现系统的精确校准。最后，在术中配准期间应用运动补偿以减少患者生理运动引起的误差。通过体模、志愿者和临床实验验证了该系统的准确性和可行性。结果表明，我们提出的自动术前表面特征识别、术中配准和定位方法为临床病变定位提供了极具潜力的解决方案。本研究通过为胸部病变实施创新且稳健的定位方法，推进了胸外科机器人系统现有的导航和临床应用方案。该集成系统在临床环境中成功得到验证，证明了其通过同时最小化辐射暴露和并发症来改变微创胸外科手术的潜力。
