冠脉造影图像三维重建的研究与实现

发布时间：2018-02-24 02:30

本文关键词： 冠脉造影三维重建 DICOM 模糊聚类蚁群算法外极线匹配　出处：《吉林大学》2017年硕士论文　论文类型：学位论文

【摘要】：医学是联系一个国家国民身心健康的核心学科,体现着国家的基础生活标准,标志着国家的综合国力。而近些年随着日常饮食质量提升和人均寿命增加,我国正在进入老龄化阶段,心血管疾病已成为第一慢性疾病,影响千万家庭的幸福,其研究工作迎来巨大挑战。冠脉造影是治疗心血管疾病的重要手段,它的原理是通过向人体的冠状动脉中注入一种无害的造影剂,这种造影剂可以在特殊的观察设备中清晰显影。冠脉造影是诊断心血管疾病的重要手段之一,无论是在安全性还是在成像效果上,都已经成为临床上的“金标准”。冠脉造影的三维重建是一个跨学科的研究领域,是医疗影像可视化研究的关键,它将得到的造影注入图像序列,以人体组织和构造的形式立体的呈现出来,从而辅助医疗人员全方位,多角度的观察病灶,有效便捷的进行疾病诊断。冠脉造影图像因为其本身成像的特殊性存在着分析上的困难。首先需要自主实现针对DICOM格式文件的解析与读取,接着由于血管组织较为细微,不易观察,分割后图像与中心线的误差影响到后期重建,需要设计并实现一种针对其特殊性的算法,最后针对二维点数据,如何运用几何变换的关系得到三维点坐标也是一大难题。本次研究将立足于现有技术,在掌握冠脉造影的成像原理和图像基本参数的基础上,首先学习DICOM3.0标准,了解其格式与各部分代表的含义,完成图像的解析与读取。其次学习并实现多种分割算法,通过调整参数来比较这些算法的优劣,为之后实现基于模糊聚类的蚁群算法打下基础。接着针对冠脉造影图像的精准分割,提出并实现基于模糊聚类的蚁群算法,最后利用数学关系推导二维点到三维点的转换,完成基于外极线匹配原理的血管段和特征点匹配。最终完成图像的三维重建。冠脉造影图像的三维重建技术能够在计算机辅助诊断方面提供更细致的图像分析,描述更精确地血管形态,从而为疾病的诊断提供更有效的数据基础和诊断信息。在医学教学方面,可以帮助缺乏经验的医生或学生全面了解血管的构造和特征,从而更有效的学习医学知识。
[Abstract]:Medicine is a core subject to contact a national mental health, reflects the country's basic living standards, marking the country's comprehensive national strength. And in recent years with the diet quality improvement and increase life expectancy, China is entering the aging stage, cardiovascular disease has become the chronic disease affecting millions of family happiness, the study work to usher in a huge challenge. Coronary angiography is an important means for the treatment of cardiovascular disease, its principle is through the injection of a contrast agent harmless to the human body in the coronary arteries, the contrast agent can be observed in the special equipment in clear imaging. Coronary angiography is one of the important means for the diagnosis of cardiovascular diseases, both in safety or in the imaging effect, has become clinically "gold standard". The 3D reconstruction of coronary angiography is an interdisciplinary research field, is the medical image The key of visualization research, it will be injected into the image sequence obtained by contrast, human tissue and structure form of stereo picture, to assist medical personnel all-round, multi angle observation of lesions, effective and convenient to diagnose the disease. Coronary angiography image because of the characteristics of the imaging analysis on difficulties exist. First need to realize to read DICOM file format parsing, then the vascular tissue is more subtle, not easy to observe, segmentation error image and the center line of the influence to the later reconstruction, need to design and implement an algorithm according to its particularity, the two-dimensional point data, the relationship between how to use geometric transform from 3D point coordinate is a major problem. This study will be based on the existing technology, based on mastering the coronary angiography imaging principle and the basic parameters of the image, first of all to learn DICOM3.0 Understand the format and standard, each part of the meaning of image analysis and reading. Secondly, learn and achieve a variety of segmentation algorithms to compare these algorithms by adjusting the parameters of the pros and cons of to lay the foundation for the achievement of the ant colony algorithm based on fuzzy clustering. Then the accurate segmentation coronary angiographic images, propose and realize ant colony the algorithm based on fuzzy clustering, finally using mathematical relations of 2D point conversion to 3D points, matching feature points and segments of epipolar line matching. Based on the principle of complete 3D reconstruction image. 3D reconstruction images of coronary angiography can provide more detailed image analysis in computer aided diagnosis, description precise vascular morphology, so as to provide basic data and diagnostic information is more effective for the diagnosis of the disease. In the medical teaching, can help the inexperienced doctors Or students can fully understand the structure and characteristics of blood vessels, so that they can learn medical knowledge more effectively.

【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R816.2;TP391.41

【相似文献】