CASIA OpenIR  > 毕业生  > 硕士学位论文
手术机器人自主颅骨铣削运动控制研究
钱琛
2024-05-16
Pages100
Subtype硕士
Abstract

开颅手术是治疗颅脑创伤、脑肿瘤、颅内出血等疾病的重要方式。传统的开颅手术方法存在手术区域准确性不高、容易损伤硬脑膜、手术过程耗时耗力、全国医疗资源分布不均匀等一系列问题。相对于传统的手术方式,手术机器人技术具备精准性高、稳定性好等优点,同时能够缩短手术时间、降低医生的工作强度,从而提高手术效率。因此面向神经外科临床应用的开颅手术机器人控制技术研究具有显著的社会效益和临床应用价值。本文在国家自然科学基金委员会、国家重大科研仪器研制项目“面向神经外科肿瘤切除的高灵敏性智能显微精准导航操作仪”(62027813)的支持下,以提高神经外科机器人颅骨铣削的智能化程度为出发点,研究机器人辅助开颅手术中颅骨铣削的运动控制问题。论文的主要内容如下:

(1)针对如何将医生拟定的铣削轨迹传递给机械臂及颅骨铣削时光学导航仪视野遮挡问题,本文提出了一种颅骨铣削路径人机交互方法与术中器械跟踪方法,并搭建了开颅手术机器人系统进行实验验证。首先,本文搭建了包括七自由度机器人模块、颅骨铣削动力学模块、力传感模块与导航模块的开颅手术机器人实验平台。随后,在该实验平台的基础上提出了颅骨铣削路径人机交互方法,完成病人、机器人和医学图像坐标系的配准工作,将医生在颅骨模型确定的铣削轨迹点位置信息和法线信息传递给机械臂。同时引入了一种双重器械定位跟踪方法,建立导航仪坐标系到工具坐标系的空间转换关系,以获取导航仪视野范围内铣刀柄的相对位置信息;当铣刀柄位置超出导航仪的视野范围时,借助机械臂关节编码器并结合机器人正运动学模型与机器人基坐标系到导航仪坐标系的映射关系对铣刀柄位置进行补偿,有效地避免光学导航存在视野遮挡的问题。本研究利用颅骨模型开展了铣削实验并完成位置误差结果分析任务。实验结果显示,定位探针、定位支架和机器人的定位精度,以及图像坐标系与颅骨坐标系的配准精度均小于1mm。在颅骨模型铣削实验中,预设轨迹与没有铣削时轨迹之间的均方根误差为0.59mm,没有铣削时轨迹与实际运行轨迹之间的均方根误差为0.73mm,这些误差在手术要求的合理范围内。此外,机器人已实现在只钻一个孔的情况下完成了整个闭环轨迹的颅骨铣削任务,从而减少了开颅手术对患者造成的潜在伤害。

(2)针对开颅手术机器人颅骨铣削过程中硬脑膜保护成功率不高及机器人自主程度低的问题,本文提出了一种无模型自适应非线性垂直力控制算法,旨在安全地完成颅骨铣削任务。该算法通过实时反馈垂直力来调节铣刀的速度和位置,确保颅骨铣削过程中垂直力不会超出设定的阈值以保护硬脑膜。为解决开颅手术机器人对医生依赖大、自主程度不高的问题,本文设计了一种基于监测进给力变化的滑动平均窗口算法,用于检测铣刀是否突破孔洞边界。该算法通过两个连续时间序列窗口实现:第一个窗口𝑇1用于跟踪进给力的变化,通过平均值的计算来滤波和减少噪声;第二个窗口𝑇2则用于缓冲,减少突破阶段数据对第一个窗口𝑇1的影响。通过持续监测进给力数据的变化,并将其与滑动窗口的平均值进行比较,实现了对铣刀是否突破颅骨边界的自动识别。本研究利用4个离体狗头和4个离体羊头开展了72次颅骨铣削实验。实验结果显示,采用所设计的控制算法时,垂直力的最大均方根误差百分比为0.99%,突破检测的成功率为98.61%。在所有实验的过程中,未观察到任何颅内硬脑膜损伤的案例,而且在整个实验过程中机器人在很少的人为干预下完成颅骨铣削任务。

(3)针对机器人颅骨铣削多方向力控制精度不高的问题,本文提出了一种混合导纳控制算法,旨在有效地完成颅骨铣削任务。首先,采用无模型自适应非线性力控制算法,实现铣刀进给方向力的精确控制,以保护铣刀和正常骨组织。其次,将铣刀的垂直力视为铣刀保护鞘的弹性形变,并结合机械臂与未知环境的导纳数学模型,将垂直力控制问题转化为一个与位置非线性相关的PD控制问题。因此,本文提出了一种基于模糊控制和无模型自适应控制算法的混合导纳控制方法,将进给力和垂直力控制在合适的范围,并在离体羊颅骨和狗颅骨上开展了实验验证。利用机器人辅助开颅系统开展的42例离体动物颅骨铣削实验结果表明,所提出的算法在3s内将力误差百分比控制在5.0%以下,垂直力和进给力的最大均方根误差百分比分别为1.85%和1.94%。此外,实验中未观察到硬脑膜损伤的情况,同时机器人系统显示出高度的自主性。

Other Abstract

A craniotomy is an important way to treat craniocerebral trauma, brain tumors, intracranial hemorrhage, and other diseases. The traditional craniotomy method has a series of problems, such as the inaccuracy of the surgical area, easy to damage the dura mater, time-consuming and labor-intensive surgical process, and uneven distribution of medical resources across the country. Compared with traditional surgical methods, surgical robotics has the advantages of high accuracy and good stability, and at the same time, it can shorten the operation time and reduce the surgeons’s work intensity, thus improving the efficiency of surgery. Therefore, the study of craniotomy robot control technology for neurosurgery clinical application has significant social benefits and clinical application value. This paper is supported by the National Natural Science Foundation of China and the National Major Scientific Research Instrument Development Project ”Highly Sensitive Intelligent Microscopic Precise Navigation Manipulator for Neurosurgical Tumor Resection” (62027813), which takes the improvement of the degree of intelligence of neurosurgical robotic skull milling as the starting point, and studies the motion control of the robotic-assisted skull milling in the craniotomy surgery. The main contents of the paper are as follows:

(1) Aiming at the problem of how to transfer the milling trajectory designed by the surgeon to the manipulator and the occlusion of the field of view of the optical navigator during skull milling, this paper proposes a cranial milling trajectory human-machine interaction method and intraoperative instrument tracking method and constructs a craniotomy robot system for experimental verification. First, this paper constructs a craniotomy robot experimental platform including a seven-degree-of-freedom robot module, a cranial milling dynamics module, a force sensing module and a navigation module. Subsequently, a cranial milling trajectory human-machine interaction method is proposed on the basis of this experimental platform, which completes the alignment of the coordinate systems of the patient, the robot and the medical image, and transmits the positional and normal information of the milling trajectory determined by the surgeon in the cranial model to the manipulator. Meanwhile, a dual instrument localization tracking method is introduced to establish the spatial transformation relationship from the navigator coordinate system to the tool coordinate system to obtain the relative position information of the milling cutter shank within the field of the navigator’s view. When the position of the milling cutter shank exceeds the field of the navigator’s view, the position of the milling cutter shank is compensated with the help of the manipulator joint encoder and the combination of the robot’s positive kinematic model and the mapping relationship from the robot’s base coordinate system to the navigator’s coordinate system, which effectively avoids the problem of field of view obstruction in optical navigation. In this study, cranial models were used to carry out milling experiments and complete the task of analyzing the positional error results. The experimental results show that the positioning accuracy of the positioning probe, the tracking stent and the robot, as well as the registration accuracy of the image coordinate system and the skull coordinate system are all less than 1 mm, which meets the requirements of surgery. In the skull model milling experiment, the root mean square error between the predefined trajectory and the no-milling trajectory is 0.59 mm, and the root mean square error between the no-milling trajectory and the actual trajectory is 0.73 mm, which is within the reasonable range of the surgical requirements. In addition, the robot has realized the completion of milling an entire closed-loop trajectory of the skull while drilling a single hole, reducing the damage to the patient caused by craniotomy.

(2) Aiming at the problem of low success rate of dural protection and low degree of robot autonomy during robotic skull milling in craniotomy surgery, this paper proposes a model-free adaptive nonlinear vertical force control algorithm, which aims to accomplish the cranial milling task safely. The algorithm regulates the velocity and position of the milling cutter through real-time feedback of the vertical force to ensure that the vertical force does not exceed the set threshold to protect the dura mater during skull milling. To solve the problem that craniotomy robots are highly dependent on surgeons and have a low degree of autonomy, this paper designs a sliding average window algorithm based on monitoring the change of feed force to detect whether the milling cutter breaks the boundary of the hole. The algorithm is realized by two continuous time series windows: the first window 𝑇1 is used to track the change of feed force, filtering and reducing the noise through the calculation of the average value, and the second window 𝑇2 is used to buffer and reduce the impact of the breakthrough phase data on the first window 𝑇1 . Automatic detection of whether the milling cutter breaks through the skull boundary is achieved by continuously monitoring the changes in the feed force data and comparing it with the average value of the sliding window. In this study, 72 cranial milling experiments were carried out using four in vitro dog skulls and four in vitro goat skulls. The experimental results showed that the maximum root mean square error percentage of vertical force was 0.99% and the success rate of breakthrough detection was 98.61% when using the proposed control algorithm. No intracranial dural damage was observed during all the experiments, and the robot accomplished the cranial milling task with little human intervention throughout the experiments.

(3) Aiming at the problem of low accuracy of multi-directional force control for robotic skull milling, this paper proposes a hybrid admittance control algorithm aimed at effectively accomplishing the task of skull milling. First, a model-free adaptive nonlinear force control algorithm is used to realize the precise control of the force in the milling cutter feed direction to protect the milling cutter and normal bone tissue. Second, the vertical force of the milling cutter is regarded as the elastic deformation of the protective sheath of the cutter, and combined with the admittance mathematical model of the manipulator and the unknown environment, the vertical force control problem is transformed into a PD control problem related to position nonlinearity. Therefore, a hybrid admittance control method based on fuzzy control and model-free adaptive control algorithms is proposed to control the feed force and vertical force within a suitable range, and experimental validation is carried out on in vitro dog skulls and in vitro goat skulls. The results of 42 in vitro animal cranial milling experiments carried out using a robot-assisted craniotomy system showed that the proposed algorithm controlled the force error percentage to less than 5.0% in 3 s, and the maximum root mean square error percentages of vertical force and feed force were 1.85% and 1.94%, respectively. In addition, no dural damage was observed in the experiment, while the robotic system showed a high degree of autonomy.

Keyword机器人辅助颅骨铣削 人机交互与术中器械跟踪 垂直力控制 突破检测 混合导纳控制
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/56603
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
钱琛. 手术机器人自主颅骨铣削运动控制研究[D],2024.
Files in This Item:
File Name/Size DocType Version Access License
钱琛_手术机器人自主颅骨铣削运动控制研究(7868KB)学位论文 限制开放CC BY-NC-SA
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[钱琛]'s Articles
Baidu academic
Similar articles in Baidu academic
[钱琛]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[钱琛]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.