显微CT三维血管图像定量分析的研究
发布时间：2018-09-09 14:24
【摘要】：CT成像技术自上世纪70年代问世以来取得了迅速的发展,在生物医学领域获得了广泛而深入的应用。同时,通过图像定量分析的方法分析三维CT图像中研究对象的形态和结构,也成为了生物医学研究的重要手段。许多疾病的产生和发展都表现出了血管异常与畸变的现象,例如血栓、肿瘤和肝硬化等。利用三维图像定量分析的方法分析血管结构的形态和特征,提取血管分支数、血管管径以及血管密度等定量参数研究医学问题,已成为许多疾病早期检测、诊断和研究治疗方案的必要辅助方法。近年来,随着X射线成像技术的不断进步,采集到的三维血管图像具有越来越高的分辨率,血管细节信息更为丰富。基于同步辐射的显微CT成像(Synchrotron radiation micro-computed tomography,SR-μCT)技术,具有高分辨和可实现相衬成像(Phase Contrast Imaging,PCI)的特点。相比于传统的X射线吸收成像,X射线相位衬度CT成像(Phase Contrast Computed Tomography)对低Z元素样品具有更高的密度分辨能力。例如对生物样品的成像中,由于不同生物软组织间密度差异较小,吸收成像难以分辨出不同的软组织;而通过相位衬度CT成像,利用相位信息分辨不同的软组织,可以获得较好衬度的三维图像。利用X射线的吸收特性采集生物样品的血管图像时,为了在图像中显示清晰的血管,往往需要对血管填充造影材料。与之相比,相位衬度CT成像无需借助造影剂,减少了样品制备的复杂性,并且对于采用吸收成像时因为造影剂颗粒太大而不能渗透的微小血管也能够采集到清晰的图像。采集到的三维血管图像分辨率高、细节丰富和数据量大的新特点对图像的定量分析方法也提出了更高的要求和新的挑战。针对上海光源X射线成像及生物医学应用光束线站(BL13W)对于三维图像定量分析的大量需求,本文开始了显微CT三维血管图像定量分析的研究,取得了一下几个方面的成果:1.发展出了基于血管树形结构提取血管定量参数的三维图像分析方法。该方法根据血管树状结构的特点,从根部区域遍历整个血管网络,提取血管分支、血管分叉点和血管长度等一系列定量参数。此外,与传统提取血管定量参数的方法相比,还得到了血管关于树形结构的定量信息,如血管分支所位于的树的层级和血管树延伸的最高层级深度等,为定量描述三维血管的结构和形态提供了更加丰富的分析手段。2.显微CT三维血管图像定量分析中,提取血管骨架(skeleton)是获得整个血管网络定量信息的关键步骤。通过血管骨架这种单像素宽度的线性结构,“简化”表征血管图形,能够方便的提取出描述血管结构和形状的量化参数,如长度和数目等。因此,提取的血管骨架是否足够精确直接决定了后续定量分析的可靠性和准确性。为了解决现有的骨架提取算法存在的过度细化、缺失分支和存在“毛刺”噪声等问题,本文提出了基于血管末端的骨架细化提取算法,提取到的血管骨架具有几何不变性。生成精确表征三维血管图形的骨架,保证了血管定量分析结果的准确性和可靠性。3.血管骨架的提取是一个极为耗费时间的运算过程,通过Open MP多线程技术实现血管骨架提取算法的并行化运算,使得提取血管骨架的时间缩短了一个数量级。对于1.95 GB大小的显微CT三维血管图像,使用16个线程进行并行运算时,可将运算时间由176 min缩短到13 min,显著的提高了提取血管骨架的时间效率。4.血管分析算法和骨架提取算法软件工具的开发实现,本文提出的算法作为软件模块集成在3D Slicer中运行。并简单介绍了利用软件工具分析CT血管图像的流程。5.应用本文三维血管图像定量分析的算法对小鼠肝脏纤维化模型进行研究。利用图像分析的方法提出了新的定量分析指标高层级血管率(HVR)用以评估不同小鼠肝脏的纤维化程度,图像分析的结果与实际血管增生现象和肝脏纤维化程度关系相符合。
[Abstract]:CT imaging technology has made rapid progress since its appearance in the 1970s, and has been widely and deeply applied in the field of biomedicine. At the same time, it has become an important means of biomedical research to analyze the shape and structure of the object of study in three-dimensional CT images by image quantitative analysis. Vascular abnormalities and distortions, such as thrombus, tumor and cirrhosis, have been demonstrated. Quantitative analysis of the morphology and characteristics of vascular structures, extraction of quantitative parameters such as vessel branch number, vessel diameter and vessel density, has become an early detection, diagnosis and treatment of many diseases. In recent years, with the development of X-ray imaging technology, the three-dimensional vascular images acquired have higher and higher resolution and more detailed information of blood vessels. Synchrotron radiation micro-computed tomography (SR-uCT) technology has high resolution and realizable phase. Compared with conventional X-ray absorption imaging, Phase Contrast Computed Tomography (PCI) has higher density resolution for low Z element samples. For example, in imaging biological samples, absorption imaging is difficult because of the small density difference between different biological soft tissues. In order to distinguish different soft tissues, three-dimensional images with better contrast can be obtained by phase contrast CT imaging, which uses phase information to distinguish different soft tissues. Compared with phase contrast CT imaging, phase contrast CT imaging does not require the use of contrast agents, which reduces the complexity of sample preparation, and can also capture clear images of small vessels that are too large to penetrate because of the size of contrast agents. Quantitative analysis methods also put forward higher requirements and new challenges. In view of the large demand for quantitative analysis of three-dimensional images by Shanghai Light Source X-ray Imaging and Biomedical Applied Beam Line Station (BL13W), this paper began the study of quantitative analysis of three-dimensional vascular images by micro-CT, and achieved the following results: 1. Based on the development of a new method of quantitative analysis of three-dimensional vascular images. According to the characteristics of vascular dendritic structure, this method traverses the whole vascular network from the root region, and extracts a series of quantitative parameters, such as vascular branches, vascular bifurcation points and vascular length. Quantitative information about the tree-like structure of the vessels, such as the hierarchy of the branches and the highest depth of the extension of the vascular tree, provides a more abundant analytical means for quantitative description of the structure and morphology of three-dimensional blood vessels. Quantitative parameters, such as length and number, can be extracted conveniently by simplifying the vascular skeleton, which is a linear structure with a single pixel width. Therefore, the reliability of subsequent quantitative analysis depends on whether the extracted vascular skeleton is accurate enough. In order to solve the problems of over-thinning, missing branches and "burr" noise in existing skeleton extraction algorithms, this paper proposes a skeleton thinning extraction algorithm based on the end of blood vessel. The skeleton extracted from the skeleton has geometric invariance. Accuracy and reliability of the analysis results. 3. Extraction of vascular skeleton is a time-consuming process. Parallel operation of the algorithm is realized by Open MP multithreading technology, which shortens the time of extracting vascular skeleton by an order of magnitude. For 3-D vascular images of 1.95 GB, 16 are used. The parallel operation of threads can shorten the operation time from 176 minutes to 13 minutes, which greatly improves the time efficiency of extracting vascular skeleton. The flow chart of CT angiography was analyzed. 5. The model of liver fibrosis in mice was studied by using the algorithm of quantitative analysis of three-dimensional angiography. A new quantitative analysis index, high-level vascular rate (HVR), was proposed to evaluate the degree of liver fibrosis in different mice, the results of image analysis and the actual proliferation of blood vessels. The phenomenon is consistent with the degree of liver fibrosis.
【学位授予单位】：中国科学院研究生院(上海应用物理研究所)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.41
本文编号：2232645
[Abstract]:CT imaging technology has made rapid progress since its appearance in the 1970s, and has been widely and deeply applied in the field of biomedicine. At the same time, it has become an important means of biomedical research to analyze the shape and structure of the object of study in three-dimensional CT images by image quantitative analysis. Vascular abnormalities and distortions, such as thrombus, tumor and cirrhosis, have been demonstrated. Quantitative analysis of the morphology and characteristics of vascular structures, extraction of quantitative parameters such as vessel branch number, vessel diameter and vessel density, has become an early detection, diagnosis and treatment of many diseases. In recent years, with the development of X-ray imaging technology, the three-dimensional vascular images acquired have higher and higher resolution and more detailed information of blood vessels. Synchrotron radiation micro-computed tomography (SR-uCT) technology has high resolution and realizable phase. Compared with conventional X-ray absorption imaging, Phase Contrast Computed Tomography (PCI) has higher density resolution for low Z element samples. For example, in imaging biological samples, absorption imaging is difficult because of the small density difference between different biological soft tissues. In order to distinguish different soft tissues, three-dimensional images with better contrast can be obtained by phase contrast CT imaging, which uses phase information to distinguish different soft tissues. Compared with phase contrast CT imaging, phase contrast CT imaging does not require the use of contrast agents, which reduces the complexity of sample preparation, and can also capture clear images of small vessels that are too large to penetrate because of the size of contrast agents. Quantitative analysis methods also put forward higher requirements and new challenges. In view of the large demand for quantitative analysis of three-dimensional images by Shanghai Light Source X-ray Imaging and Biomedical Applied Beam Line Station (BL13W), this paper began the study of quantitative analysis of three-dimensional vascular images by micro-CT, and achieved the following results: 1. Based on the development of a new method of quantitative analysis of three-dimensional vascular images. According to the characteristics of vascular dendritic structure, this method traverses the whole vascular network from the root region, and extracts a series of quantitative parameters, such as vascular branches, vascular bifurcation points and vascular length. Quantitative information about the tree-like structure of the vessels, such as the hierarchy of the branches and the highest depth of the extension of the vascular tree, provides a more abundant analytical means for quantitative description of the structure and morphology of three-dimensional blood vessels. Quantitative parameters, such as length and number, can be extracted conveniently by simplifying the vascular skeleton, which is a linear structure with a single pixel width. Therefore, the reliability of subsequent quantitative analysis depends on whether the extracted vascular skeleton is accurate enough. In order to solve the problems of over-thinning, missing branches and "burr" noise in existing skeleton extraction algorithms, this paper proposes a skeleton thinning extraction algorithm based on the end of blood vessel. The skeleton extracted from the skeleton has geometric invariance. Accuracy and reliability of the analysis results. 3. Extraction of vascular skeleton is a time-consuming process. Parallel operation of the algorithm is realized by Open MP multithreading technology, which shortens the time of extracting vascular skeleton by an order of magnitude. For 3-D vascular images of 1.95 GB, 16 are used. The parallel operation of threads can shorten the operation time from 176 minutes to 13 minutes, which greatly improves the time efficiency of extracting vascular skeleton. The flow chart of CT angiography was analyzed. 5. The model of liver fibrosis in mice was studied by using the algorithm of quantitative analysis of three-dimensional angiography. A new quantitative analysis index, high-level vascular rate (HVR), was proposed to evaluate the degree of liver fibrosis in different mice, the results of image analysis and the actual proliferation of blood vessels. The phenomenon is consistent with the degree of liver fibrosis.
【学位授予单位】：中国科学院研究生院(上海应用物理研究所)
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP391.41
【参考文献】
相关期刊论文 前8条
1 彭浩杰;赵俊;;血管网络定量分析研究进展[J];生命科学;2013年08期
2 张国强;周虎;和友;刘慧强;王玉丹;薛艳玲;谢红兰;肖体乔;;基于极坐标变换去除计算机层析图像环形伪影[J];光学学报;2012年05期
3 向秋静;李晶;刘苹;孙建奇;;基于同步辐射的不同时期肿瘤新生血管形态结构定量的初步分析[J];中国医疗器械杂志;2012年02期
4 吕永钢;杨力;;血管网分形研究进展[J];化工学报;2010年12期
5 桂建保;胡战利;周颖;郑海荣;;高分辨显微CT技术进展[J];CT理论与应用研究;2009年02期
6 黄坤武;唐杰;武港山;;针对面片的Reeb图骨架抽取算法[J];系统仿真学报;2006年S1期
7 冼鼎昌;;同步辐射的现状和发展[J];中国科学基金;2005年06期
8 庞彦伟,王召巴,林敏,岳海霞;CT图像环形伪迹分析[J];华北工学院学报;2001年01期
，本文编号：2232645
本文链接：https://www.wllwen.com/shoufeilunwen/xxkjbs/2232645.html
