脊柱微创手术机器人跟踪和导航方法研究
文献类型：学位论文
| 作者 | 宋国立
|
| 学位类别 | 博士 |
| 答辩日期 | 2016-05-30 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 韩建达 |
| 关键词 | 手术机器人 扩展集员滤波 图像配准 目标定位 |
| 其他题名 | Research on Tracking and Navigation Method of Minimally Invasive Spine Surgery Robot |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 随着手术机器人应用范围的日益扩展，在复杂手术环境下，手术规划轨迹传送到手术机器人以及手术机器人高精度、高稳定性地执行手术规划成为了手术机器人研究的重点和难点。本文以国家自然科学基金重点项目“脊柱微创智能手术环境基础问题研究”为依托，以中国科学院沈阳自动化研究所自主研发的智能手术环境为实验平台，针对手术机器人在复杂手术环境中的手术轨迹规划、手术机器人定位和导航等问题，展开深入研究，旨在提高手术机器人复杂环境下的精确性和鲁棒性。本论文的主要内容如下：首先，阐述了论文的研究背景和目前手术机器人的发展现状，并提出了手术机器人的层次划分和关键问题。依据手术机器人目前的关键问题介绍医学图像配准方法和手术机器人定位导航方法的相关研究，归纳出手术机器人亟待解决的科学问题。第二，基于特征的异维多模态医学图像配准优化算法。针对异维多模态医学图像配准问题，首先介绍了图像模态和维数转换，并介绍了图像识别和配准中常用的特征。在建立图像配准状态空间方程后，介绍了基于扩展集员滤波算法（ESMF）图像配准方法，并与基于EKF和UKF配准优化算法进行比较，分析其优点和缺陷，针对脊柱图像实验验证了算法的有效性和可行性。第三，基于扩展集员滤波算法的动态机器人定位方法。针对动态机器人定位问题，在手术环境的复杂性和手术安全性的需求下，结合手术中可能存在的实际问题，介绍了基于ESMF算法的单传感器目标定位和基于ESMF算法的多传感器动态目标合作定位两种方法。进一步根据ESMF的目标定位方法，设计了机器人控制输入。仿真实验验证了这两种算法的可行性，同时针对两种算法进行了比较，合作定位结果对单机的定位效果有显著的改善。结合ESMF的目标定位方法的机器人的控制方法，其控制精度有显著提高。第四，基于速度场的手术机器人高精度实时轨迹规划。速度场控制较传统的轨迹定位方式在轨迹定位误差和轨迹可控可预判性上有着明显的优势。在时间轨迹跟踪条件下，手术机器人在扰动下会对脊柱神经造成损伤。同时，速度场建立手术机器人空间和速度的映射，准确的位置可以获取更精确的速度，结合ESMF定位算法的速度场控制可以提高手术机器人的控制精度。为了进一步的保证脊柱的安全性，控制器设计过程中建立了规划轨迹为核心的主速度场和保护脊柱神经的保护场，仿真实验验证了速度场控制算法的有效性。最后介绍手术系统平台。针对脊柱微创手术中的若干关键问题，本章提出了微创手术机器人系统，搭建手术相关操作平台，将Staubli Tx60 定义为手术机器人，同时介绍了术前CT图像和术中X射线图像配准系统，OARM图像采集系统，以及结合NDI Polaris 光学定位定位系统的手术机器人导航系统。本文建立了基于图像配准的状态空间模型，并结合ESMF算法提高了图像配准的精度，提出了基于ESMF目标定位方法，通过单传感器和多传感器合作定位，抑制了噪声对定位过程造成的影响，提高了定位的精度和稳定性。提出了基于速度场的手术机器人高精度实时轨迹规划，在高精度定位的基础上结合速度场控制算法，可以实现手术机器人轨迹精确控制和平滑。 |
| 英文摘要 | With the growing scope of applications for surgical robots, the transfer of planned surgical activities to surgical robots in complex surgical environments, along with the high precision and high stability of the performance of surgical robots in surgical planning, have become crucial and difficult points in the research of surgical robots. Supported by the “Fundamental Problem Study on Minimally Invasive Spine Surgery in an Intelligent Surgery Environment”, a key project of the National Natural Science Foundation of China, this paper adopted an intelligent surgical environment developed by the Shenyang Institution of Automation, Chinese Academy of Sciences, as a test platform, and conducted in-depth studies focusing on the surgical planning, positioning, and navigation issues of a surgical robot in a complex surgical environment. The goal of this paper was to improve the precision and robustness of surgical robots performing in complex environments. The main content of this paper is as follows. First, the research background of this paper and the current status of surgical robot development were presented, and the hierarchical division and key issues of surgical robots were proposed. To focus on the existing key issues of surgical robots, related research on medical image registration methods and surgical robot navigation methods was introduced, and the scientific issues of surgical robots to be solved were summarized. Second, this paper presented a feature-based different-dimensional multimodal medical image registration optimization algorithm. With regard to different-dimensional multimodal medical image registration, image modality and dimension conversion were first introduced, followed by the commonly used features in image recognition and registration. After establishing an image-registration state space equation, the image registration method based on the extended set-membership filter (ESMF) algorithm was introduced and compared with the EKF- and UKF-based registration optimization algorithms to analyze the advantages and disadvantages of ESMF. The feasibility and effectiveness of the ESMF algorithm were validated in a spine-imaging experiment. Third, a dynamic robot-positioning method based on the ESMF algorithm was described. With regard to dynamic robot positioning, under the requirements of the surgical environment and surgical safety, two methods were presented that combined practical problems that may exist during surgery: the single-sensor target positioning method and the multisensor dynamic target copositioning method. Both methods are based on the ESMF algorithm. The robot control input was further designed according to the ESMF targeting method. The simulation results verified the feasibility of these two algorithms. Comparing the results of the two algorithms showed that the copositioning method was significantly superior. The robot control algorithm, combined with the ESMF target positioning method, significantly improved control precision. Fourth, the high-precision, real-time trajectory planning of a surgical robot based on a velocity field was presented. Compared with traditional trajectory-tracking methods, velocity field control has obvious advantages in minimizing trajectory-tracking errors, better controllability, and more accurate predictability. Under time trajectory tracking conditions, a surgical robot can cause spinal nerve damage owing to perturbations. On the other hand, a velocity field would establish a space-velocity mapping of the surgical robot so that accurate positioning can lead to a more accurate speed. Velocity field control integrated with the ESMF positioning algorithm can improve the control precision of a surgical robot. During the controller design process, a main velocity field from trajectory planning, and an auxiliary protection field to protect the spinal nerves, were built to further ensure safety at the spine. The simulation results showed the effectiveness of the velocity field control algorithm. Finally, a surgery system platform was introduced. Focusing on several key issues in minimally invasive spine surgery, this section proposed a minimally invasive surgical robot system, and constructed a surgery-related operating platform. Staubli Tx60 was defined as the surgical robot in the study. Meanwhile, the OARM image acquisition system (an image registration system for preoperative CT images and intraoperative X-ray images) and a surgical robot navigation system, combined with the NDI Polaris optical positioning system, were introduced. This paper established a state space model based on image registration, improved the precision of image registration by combining it with the ESMF algorithm, proposed a targeting method based on ESMF, adopted single-sensor and multisensor copositioning, suppressed the effect of noise on the positioning process, and enhanced the precision and stability of the positioning. The high-precision, real-time trajectory planning of a surgical robot based on a velocity field was presented. This system integrated a velocity control algorithm on the basis of high-precision positioning, and can be implemented to realize precise and smooth control of the trajectory of a surgical robot. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 119页 |
| 源URL | [http://ir.sia.cn/handle/173321/19676]  |
| 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 | 宋国立. 脊柱微创手术机器人跟踪和导航方法研究[D]. 中国科学院沈阳自动化研究所. 2016. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。