毕业设计-医学图像配准系统的研究与分析.docVIP

医学图像配准系统的研究与分析 摘 要 医学图像配准技术是目前医学图像处理中的研究热点，具有重要的理论研究和临床应用价值，如在病灶定位、指导神经手术、放射治疗计划以及检查治疗效果上，其目的是帮助医生提高诊断、监护和治疗水平。 本设计通过对医学图像配准的有关背景、原理及方法进行研究，采用Matlab来编程开发一个医学图像配准系统，从而实现对于人体相同部位而具有不同特征的两种不同类型的医学图像进行配准。 图像配准有两个重要环节。一是基于图像灰度的方法主要采用互信息，互信息表述了图像像素灰度信息的统计特性。基于互信息的图像配准方法具有不需要对图像预处理、自动化程度高、配准精度高等优点。在这里选用MI、EMI、GMI、RMI等互信息方法。 二是图像配准的优化算法。在本设计中主要选用了粒子群PSO和Powell优化算法，粒子群PSO主要是通过迭代法寻求全局最优解最终达到配准，而Powell主要是直接搜索求最值最终达到配准。配准过程中首先提取图像的轮廓边界，然后选取算法计算参数进行空间变换，最终实现了对两幅不同类型的医学图像进行配准的目的。 关键词：图像配准，优化算法，特征提取，灰度变换，互信息 THE ANALYSIS AND RESEARCH OF MEDICAL IMAGE REGISTRATION ABSTRACT Medical image registration is a highlight of current research on medical image processing. It has important effect on clinical diagnoses and therapies, such as on the orientation of the disease, the plan of the radiation therapy, guiding the nervous surgery and examining the therapeutic effect. Its purpose is to help doctors improve the level of diagnose, guidance and therapy. Studying the background, principle and methods, I’ll develop a system of medical image registration by Matlab in order to register two different types of medical images. Image registration has two important aspects. The intensity-based methods selects mutual Information, which shows the statistical characteristics. Because image registration based on mutual information is of not required for image processing, high automatization and high accuracy in registration. Here it can choose MI, EMI, GMI, and RMI and so on. Second, the optimization algorithm of image registration. In this design, we can main use either PSO or Powell algorithms to register images, PSO mainly through the iterative method for global optimal solutions ultimately achieve registration, but Powell mainly direct search minimum ultimately achieve registration. First, the contour of images should be obtain in the process of image registration, and then select optimization algorithms to transform, finally it succeeds in registering two different types o