机器人辅助胃镜介入关键技术研究

CORC > 沈阳自动化研究所 > 中国科学院沈阳自动化研究所 > 机器人学研究室

题名	机器人辅助胃镜介入关键技术研究
作者	李言民
学位类别	博士
答辩日期	2016-11-28
授予单位	中国科学院沈阳自动化研究所
导师	韩建达 ; 刘浩
关键词	胃镜临界滑移力周向夹持建模自主导航
其他题名	Research on Key Technologies of Robot-assisted System for Gastroscope Intervention
学位专业	模式识别与智能系统
中文摘要	胃镜是诊断治疗上消化道疾病的主要工具。目前随着微创手术的发展，以胃镜为基础开展的镜下手术以其创伤小、恢复快的优势得到了广泛的应用。然而，胃镜检查在临床应用中尚存在一些问题亟待解决，第一，胃镜操作对医生要求高，需要医生进行专门的培训并且培训周期长，个体操作误差大；第二，国内消化道疾病患者众多，而训练有素的医生严重不足，导致医生工作强度特别大；第三，部分胃镜介入治疗手术使用X射线图像进行引导，射线辐射以及胃肠液污染威胁着医生的身心健康，也使工作效率大大下降。为解决上述问题，本文在“十二五国家科技支撑计划”课题的资助下，研制了胃镜辅助介入机器人系统，并深入研究了机器人辅助胃镜介入过程中的镜体输送、建模与自主导航三项关键技术。研制了胃镜辅助介入机器人系统。借鉴胃镜手工操作特点，采用双臂构型。镜体输送臂调整胃镜相对病人口腔的位姿，实现持镜与送镜动作；对输送臂进行了逆运动学分析，确定各关节参数与基座位置；分析输送器与镜体间的作用力确定电机输出力矩。镜体操作机械臂控制操作部完成镜体旋转、操作部跟随、末端弯曲功能。使用电气化方式实现对送气送水和抽吸功能的控制。检测胃镜介入长度，实现双臂协同操作，使镜体处于自然受力状态，减少镜体介入长度损失，保障操作安全与力矩的有效传递。分析主从操作手柄映射方案，实现胃镜的直观操作。最后，与同类机器人比较，说明本文提出机器人系统的先进性。基于临界滑移力提出了气囊式周向均匀夹持机构实现胃镜的输送。与传统摩擦轮机构相比，本方法可通过改变气压实时调节输送机构的临界滑移力，通过镜体在机构内滑移使镜体组织交互力维持在安全的范围。此外，该方法以面接触提供更大的接触面积，在相对小且均匀的夹持压强下具有足够大的输送能力，可在输送过程中更好地保护胃镜。通过对气囊剪切刚度进行分析，优化气囊参数，在保证功能的前提下，降低气囊弹性对输送精度的影响。采用模糊PID控制改善系统的气路响应特性。本文提出的气囊式周向均匀夹持输送机构可根据介入位置实时调节气路气压，从而保证介入操作的安全性。从力学上对胃镜弯曲段非线性特性进行了分析建模。首先，对胃镜内部蛇骨、编网、橡皮的结构进行剖析，使用D-H参数方法进行运动学建模，获取胃镜弯曲段的工作空间；其次，引入库伦摩擦，采用迭代的方式对钢丝沿弯曲段的拉力分布进行分析，获取各关节的弯曲角度。实验结果表明，摩擦力是引起弯曲段非线性特性的重要因素。通过精确建模可提升系统的控制精度。通过人机协作任务分析，将镜下靶点跟踪交由机器人自主完成，进退镜动作由人通过手柄控制。以胃镜自主介入为例进行研究，提出了基于平滑间隙逆补偿的胃镜自主导航方法。首先，通过滤波与对比度拉伸变换对图像进行预处理，采用暗区识别与边缘检测的方法获取腔道中心；其次，对间隙与间隙逆模型进行分析，并分别测试获得胃镜上下弯曲与左右弯曲的非线性参数；最后，为防止间隙逆模型在输入速度为零时的频繁抖动，采用二阶低通滤波器进行处理，获取平滑间隙逆对系统进行补偿控制。自主导航算法的实现，可避免医生对胃镜拨轮的频繁操作，提升操作的舒适性。最后搭建了胃镜辅助介入机器人系统样机实验平台，介绍了样机控制系统的硬件与软件构成。首先开展气囊式周向均匀夹持输送机构人体模型实验与动物实验，证实了机构的有效性，并验证了系统的性能。其次开展主从操作方式动物临床实验，验证主从式操作方式的有效性与性能。结果发现，机器人辅助介入提升了操作的直观性，降低了操作的难度，但需要操作者多次训练对系统熟悉提升操作效率。最后，开展了自主介入人体模型实验，验证了自主导航算法的有效性。实验结果表明，算法可有效引导腔道中心向视野中心移动。通过与人手介入结果比较，发现自主导航算法在提升操作舒适性的同时，提升了操作效率。综上所述，本文研制了胃镜辅助介入机器人系统，并对柔性镜体输送、非线性模型、自主导航等关键技术进行了深入的研究，最后开展实验对系统与算法性能进行验证，这些研究为胃镜辅助介入机器人的临床应用奠定了基础。在胃镜辅助介入机器人基础上进行手术功能扩展是下一步的研究方向。
英文摘要	Gastroscope is the main tool for diagnosis and therapy of upper digestive tract diseases. With the development of minimally invasive surgery, endoscopic surgery based on gastroscope has been widely used owing to minimal damage and fast recovery. However, some problems during clinical application need to be solved. Firstly, gastroscope manipulation is difficult to be mastered. Even though the clinicians have accepted special and long-time training, the operation level of them is still different. Secondly, large number of patients and the lack of skillful clinicians increase the labor intensity of endoscopists greatly. Thirdly, some endoscopic surgeries use X-ray as visual feedback. The radiation and digestive juices pollution threatens the health and decreases the working efficiency of clinicians. To solve these problems, this paper designs a robot-assisted system for gastroscope intervention and focuses on the key technologies of the system included shaft delivery, nonlinear modeling and autonomous navigation supported by the National Key Technology R&D Program for the 12th five-year plan. This paper developed a robot-assisted system for gastroscope intervention. The system is divided into two parts - delivery robot arm and operation robot arm – based on the sufficient analysis of characteristics of manual manipulation. The delivery robot arm adjusts the pose of gastroscope relative to patient mouth, clamps the gastroscope and implements insertion. Inverse kinematic analysis of the delivery robot arm was constructed to determine the joint parameters and the position of base. The force between gastroscope and delivery device was also analyzed to determine the output torque of motor. The operation robot arm realizes the rotation, follow of control section and distal bending. Electrified method is used to realize gas supply, water supply and suction. Synergetic operation of two arms was realized based on the detected gastroscope insertion length to decrease the lack of insertion length and guarantee safe operation and effective torque transmission. The projection of the DOF of robots with controlling joystick for master-slave operation was analyzed to realize intuitive operation of gastroscope. Finally, the advancement of our system was illustrated by comparing with homogeneous robot system. This paper proposed a pneumatic circumferentially clamped mechanism for gastroscope delivery based on critical slipping force. Compared with friction wheels, this method could adjust the critical slipping force real-time by changing air pressure and keep the interaction force between gastroscope and tissue in safe range by slipping of gastroscope. Besides, it has much bigger contact area and provides enough delivery force with small clamp pressure to protect gastroscope better. Then, the shear stiffness of gasbag was analyzed to optimize the parameters and minimize the influence of deformation to delivery accuracy. Besides, fuzzy PID controller was used to improve the response of pneumatic system. The presented pneumatic circumferentially clamped mechanism could guarantee the operation safety by adjusting the air pressure in time according to the intervention position. The nonlinear characteristics of gastroscope bending section was analyzed and modeled in mechanics. Firstly, the structures of gastroscope including snake bone, net and rubber tube was dissected. Kinematic model was constructed by D-H parameter method. The workspace was also calculated. Then, Coulomb friction was used to analyze the tension force distribution along the cable by iteration. The bending angle of each joint was calculated. The results showed that the friction force is the main cause of nonlinear characteristics of gastroscope bending section. The precise model of gastroscope bending section could improve the control accuracy of the robot system. According to task analysis of human-robot collaboration, the tracking to endoscopic target was allotted to the autonomous robot and the delivery of gastroscope was controlled by joystick. Aiming at autonomous gastroscope intervention, this paper proposed an autonomous navigation method based on smooth backlash inverse. Firstly, Gaussian filter and contrast stretching was used to preprocess the endoscopic images. Dark region recognition and edge detection was used to get the lumen center. Then, the backlash and backlash inverse model was analyzed. Nonlinear parameters of up/down and left/right bending was measured. Finally, smooth backlash inverse using second order low pass filter was used to compensate the nonlinearity. It aims to avoid the continuous switching when input velocity is close to zero. The realization of autonomous navigation keeps the clinicians out from the frequent operation of gastroscope knobs and increase the operation comfort. Finally, the experimental platform of robot-assisted system for gastroscope intervention was constructed. The hardware and software of the controller was introduced. Firstly, in vitro and in vivo experiment were performed onto the pneumatic circumferentially clamped mechanism for gastroscope delivery, the results verified the efficiency and performance of the system. Then, in vivo animal experiment was performed to verify the validity and performance of master-slave operating mode. The results showed that the robot improved the operating intuition and decrease the manipulation difficulty. The operator should accept more trainings to improve operation efficiency. At last, autonomous navigation algorithm was performed on human phantom. The results indicated that the algorithm could guide the lumen center moving to center of view. Compared with manual operation, the autonomous navigation algorithm could improve operation comfort and increase efficiency. In conclusion, this paper researches on the robot-assisted system for gastroscope intervention and key technologies such as flexible endoscope delivery, nonlinear model and autonomous navigation. Experiments were done to verify the performance of the proposed system and algorithm. This research lays the foundation for the clinician application of the system. The extension of robot surgery based on this system is future direction in the research.
语种	中文
产权排序	1
页码	113页
内容类型	学位论文
源URL	[http://ir.sia.cn/handle/173321/19458]
专题	沈阳自动化研究所_机器人学研究室
推荐引用方式 GB/T 7714	李言民. 机器人辅助胃镜介入关键技术研究[D]. 中国科学院沈阳自动化研究所. 2016.