基于冠状动脉CT造影的冠心病辅助诊断方法研究

发布时间：2018-06-18 00:04

本文选题：冠心病 + 计算机层析成像　；参考：《天津大学》2016年硕士论文

【摘要】：冠心病被视为威胁人类健康的主要因素之一。临床研究表明,64排螺旋CT对于冠心病的筛查具有与传统冠状动脉X射线造影相媲美的精度。近年来,基于斑块衰减模式的冠状动脉病变新型分类标准的提出,对冠状动脉病变诊察提出了更高的要求。本文对基于冠状动脉CT造影的冠心病辅助诊断算法进行了研究。该算法依据新型分类标准能够实现对冠状动脉病变的自动分类,为医生提供可供参考的辅助诊断信息,提升了冠心病诊断精度和效率。除此之外,本文利用开源软件开发工具包VTK完成了医学图像交互显示软件的设计,实现了基本图像可视化功能。本文主要工作如下:1.设计了基于冠状动脉CT造影的冠心病辅助诊断算法。该算法实现的技术流程包括冠状动脉血管结构模型建立和冠状动脉血管病变探测和分类两部分。其中,冠状动脉血管结构模型建立包括了图像预处理、血管中心线提取和血管腔壁提取三部分。2.设计了基于各向异性扩散滤波和基于Hessian矩阵的血管滤波器的图像预处理方法。该方法能够在抑制噪声的同时在一定程度上提升图像中血管的对比度,其处理结果能够保证后续中心线提取算法的输入有效。3.采用基于最小代价路径的血管中心线提取算法完成冠状动脉中心线提取,提出了基于邻域特征和灰度特征的启发式算法策略对传统算法进行了改进。该改进方法能够在保证提取精度不变的同时,降低算法运行的时间和空间开销。4.利用血管中心线作为源点,设计一维投射“光线”进行图像采样,基于血管腔壁在图像上的灰度衰减特性,获取“光线”上的腔壁位置,之后通过轮廓拟合获取血管腔壁,建立血管结构模型,相对于传统算法提升了提取精度。5.利用开源软件开发包VTK完成了医学图像数据可视化功能开发,实现了DICOM图像浏览、图像多平面重建等功能,搭建了医学图像可视化软件平台。
[Abstract]:Coronary heart disease is regarded as one of the main factors threatening human health. Clinical studies have shown that 64-slice spiral CT screening for coronary heart disease has a accuracy comparable to that of conventional coronary angiography. In recent years, a new classification standard of coronary artery disease based on plaque attenuation mode has been proposed, which has put forward higher requirements for the diagnosis of coronary artery disease. In this paper, coronary artery disease assistant diagnosis algorithm based on coronary CT angiography is studied. According to the new classification standard, the algorithm can realize the automatic classification of coronary artery disease, provide doctors with the reference of auxiliary diagnosis information, and improve the accuracy and efficiency of coronary heart disease diagnosis. In addition, the design of medical image interactive display software is completed by using open source software development toolkit VTK, and the basic image visualization function is realized. The main work of this paper is as follows: 1. An auxiliary diagnosis algorithm for coronary artery disease based on coronary CT angiography was designed. The implementation of the algorithm includes two parts: coronary artery structure model, coronary artery lesion detection and classification. Among them, the establishment of coronary artery structure model includes three parts: image preprocessing, centerline extraction and vascular wall extraction. An image preprocessing method based on anisotropic diffusion filter and Hessian matrix based vascular filter is designed. This method can suppress the noise and improve the contrast of the blood vessels in the image to a certain extent, and the processing result can ensure that the input of the subsequent centerline extraction algorithm is effective. 3. The coronary artery centerline extraction algorithm based on minimum cost path is used to extract the coronary centerline. A heuristic algorithm strategy based on neighborhood feature and gray feature is proposed to improve the traditional algorithm. The improved method can reduce the time and space cost of the algorithm while keeping the extraction accuracy unchanged. Using the center line of the vessel as the source point, the one-dimensional projection "ray" is designed for image sampling. Based on the gray attenuation characteristics of the vascular wall on the image, the position of the cavity wall on the "light" is obtained, and then the vascular wall is obtained by fitting the contour. Compared with the traditional algorithm, the model of vascular structure can improve the precision of extraction. The open source software package VTK is used to develop the visualization function of medical image data. The functions of DICOM image browsing and image multi-plane reconstruction are realized. The medical image visualization software platform is built.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R541.4;TP391.41

【相似文献】