【作者】 朱国昕；

【导师】 郭庆鼎； 李玉；

【作者基本信息】 沈阳工业大学 ， 医学电磁工程， 2008， 博士

【摘要】 本文为吴阶平医学基金会临床科研资助基金课题,主要研究调强适形放射治疗及电动多叶光栅质量控制,并以此为背景结合临床应用,对调强适形放射治疗系统中临床应用方面的一些关键技术进行研究,进一步完善放射治疗系统,改进临床应用中电动多叶光栅不完善的电机驱动问题。从而完善整个调强适形放射治疗系统,解决临床实际问题。主要研究内容如下:医学图像分割问题是调强适形放射治疗三维重建的基础,图像分割好坏直接影响放射治疗照射的精度。本文以自动图像分割为主研究医学图像分割问题,提出一种自适应能量函数的马尔可夫随机场模型,在能量函数最小化过程(即像素标记场的更新过程)中动态调整条件概率模型能量函数的权值,进而获得观测场的全局最优的参数。在此基础上,利用形态学算子对肿瘤图像分割结果进行形态滤波,经过开运算和闭运算去掉空洞和孤立的噪声点,在肿瘤图像分割结果的基础上引入Canny边缘提取算子对区域进行边缘化,提取出封闭的边缘。医学图像三维重建是放射治疗系统的关键,而体数据的构造是三维重建的前提,本文提出了一种基于矢量距离变换的符号距离函数计算方法,首先定义符号距离函数,然后提出符号函数的构造方法,提高符号距离场计算的速度和效率,这种方法不但精度高,鲁棒性强,还加快了体数据场的构造。在体数据构造的基础上,利用MC(Marching Cubes)方法和MT(Marching Tetrahedral)方法对体数据进行三维重构实现可视化。研究了MT算法解决MC算法在实际应用中存在的二义性问题的方法。本文分别对规则野剂量模型、不规则野剂量模型进行了分析和总结,介绍了几种经典剂量计算方法蒙特卡罗法、Clarkson散射求和方法,研究了Day氏函数法,通过对标准散射剂量数据进行扇形积分求等效方野的Day氏函数法,计算不规则野的剂量分布,达到较高的精度。电动多叶光栅是调强适形放射治疗系统的关键技术和执行部件,探讨了电动多叶光栅适形调强的实现方式,对电动多叶光栅静态调强技术研究了其剂量分布的分级方法,提出了各射野形状的确定思路和方法;对电动多叶光栅动态叶片调强技术研究了调强原理,提出了满足电动多叶光栅的叶片运动模型及其控制策略,对叶片的运动规律及有限加速度问题进行了探讨,针对电动多叶光栅中直流伺服电动机、步进电机故障多的缺点,提出了用无刷直流电动机作为控制部件,进行临床实验解决了临床实际问题。最后,对调强适形放射治疗系统中图像分割、图像三维重建等关键技术研究进行了仿真实验,对无刷直流电动机作为电动多叶光栅控制部件进行临床应用验证,验证了以上的研究工作是有效的、可行的。更多还原

【Abstract】 The project is aided financially by Wu Jieping Medical Foundation for Clinical Research. Intensity Modulated Radiation Therapy (IMRT) and the quality control of multi leaf collimator (MLC) are chiefly reaserched in this thesis. Based on that, some key techniques on clinical application of IMRT are lucubrated, the radiotherapy system is further perfected, and the matter about imperfect motor drive of MLC in clinical application is improved. The chief goal of this project is to perfect IMRT system and to solve the questions of clinical applications. The main contents are as follows.The problem about segmentation of medical images is the base of three-dimension reconstruction during IMRT, since the irradiation precision during radiotherapy is directly affected by the quality of image segmentation. In this dissertation, the research on medical image segmentation is carried out main aiming at automatic image segmentation, a MRF (Markov Random Field) model about self-reacting energy function is proposed, during the energy function is minimizated, the weighting factors of the energy function in the conditional probability model are dynamically adjusted, and the global optimization parameters about the observation field are obtained. Based on that, the segmented results of tumor images are filtered with morphology operator, some holes and isolated noise elements are eliminated by opening and closing operations. Finally, on the base of the segmented results of tumor images, Canny edge extraction operator is used to margin the regions and extract the closed edge.Three-dimension reconstruction of medical images is the key of radiation therapy system, and construction of volume data is the prerequisite of three-dimension reconstruction. In this dissertation, a method based on vector distance transformation is proposed to calculate the symbol distance function. By this method, which firstly define a symbol distance function and then put forward the construction method of the symbol function, the speed and efficiency of calculating symbol distance field are raised greatly. This method is not only precise and highly robust, but promotes the construction of volume data field. Marching cubes (MC) and marching tetrahedral (MT) algorithm are used to complete the three-dimension reconstruction of volume data and realize visualization based on the construction of volume data field. In the project, MT algorithm is probed into, and the practical application matter of MC algorithm is solved.The dose models of regular and irregula fields are analyzed and summarized in this thesis. A few traditional dose algorithms such as Monte Carlo method and Clarkson’s scatter summation method are introduced, Day’s function method is studied, by which the equivalent square field is obtained through sector integration to the standard scatter dose data and the dose distribution in irregular field is exactly calculated.The realizing ways of intensity modulation by MLC which is an important executive part of IMRT are discussed. Aiming at static intensity modulation techniques by MLC, the grade scale of dose distribution is also discussed, the methods and train of thoughts which are used to define the shapes of different irradiation fields are proposed; For dynamic leaf intensity modulation by MLC, the intensity modulation theory is studied, leaves moving model and strategy are put forward, leaves moving regular and the matter about the acceleration limit of leaves are discussed. Aiming at the weakness that the DC servo motors and step motors of MLC often have a breakdown, the brushless DC motors which have no carborn brush are proposed as control parts of MLC. By clinical experiment, the practical application questions of MLC in clinic are solved.Finally, the emulation experiments for the key techniques such as image segmentation and 3D reconstruction are carried out, the efficiency of brushless DC motors used as the control parts of MLC is put to the proof by clinical application, and the reaserches are verified to be effective and practicable.更多还原