微循环及心肌细胞显微图像特征参数动态测量技术及应用研究

【作者】 盛有明；

【导师】 修瑞娟； 张正国；

【作者基本信息】 北京协和医学院 ， 生物医学工程， 2011， 博士

【摘要】 心血管系统是维持基本生命活动的动力系统,无论心脏还是血管系统均存在节律性舒缩功能。大量研究还证明,微血管和微淋巴管也普遍存在着自主节律性的收缩/舒张运动(自律运动),这种自律运动是除心脏驱动之外,微循环系统对组织和器官灌注的能量来源之一。有关生命的节律性问题,特别是循环系统的节律性特征正在成为医学相关研究的热点问题。心肌细胞是心脏泵血功能的基本收缩单元,因此以离体培养的心肌细胞为模型,对其搏动和收缩特性进行动态分析,有助于研究心肌细胞功能,探索心血管疾病的发病机理。微血管、微淋巴管以及流动于其管腔中的体液所共同组成的微循环系统是循环系统的重要组成部分,是氧、营养物质和信号分子最主要的交换场所。微血管与微淋巴管的自律运动不但具有调节体液流动的作用,而且是“海涛式组织灌注”的重要能量来源之一。无论是微血管、微淋巴管还是心肌细胞的节律性运动都是一种具有时相变化的动态过程,目前基于数字图像研究其特征指标,均需要从一系列连续分析图像序列中的前后变化提取其特征参数。因此,快速、准确的自动化特征指标动态测量是研究上述节律性问题的技术基础。由于技术手段的限制,目前在对心肌细胞搏动、微血管和微淋巴管自律运动的研究中,尚缺乏有效的定量特征参数,或缺乏动态自跟踪检测特征参数的技术手段。本研究正是为解决本研究所在心肌细胞搏动和收缩特性以及微血管和微淋巴管自律运动特性研究中所遇到的技术瓶颈问题,通过创新提出显微数字图像处理的解决方案和软件编程,实现了特征参数基于静态图像的自动提取和基于动态视频图像的自跟踪检测,为相关研究提供有力的创新性技术支持。本研究从静态图像和动态视频两个方面,分别对微循环图像、心肌细胞搏动视频以及微血管自律运动视频中特征参数的测量方法进行了研究。在以下三个方面取得了一定的技术改进、提升或创新：(1)对部分静态微循环图像网络特征参数借助人机交互手段初步实现了自动化提取、测量以及分析计算。(2)对心肌细胞自主搏动行为,在搏动位移和频率自动检测的基础上,提出描述其收缩特性的新指标,并创新设计了线性动态模板自跟踪检测的解决方案,即“位移模板自跟踪算法”,开发出了多参数同步化动态测量的程序软件。应用最新设计的检测系统,本研究初步分析了不同心脏部位来源的心肌细胞搏动和收缩行为的力学特性参数,如搏动幅度、收缩速度、收缩加速度等,发现了不同部位来源的心肌细胞除以往描述的搏动频率的不同外,其收缩的力学特性也存在较大差异,这些差异与其所在部位的功能特性相关。(3)针对微血管和微淋巴管自律运动视频,创新设计了基于动态模板的解决方案,即“区域特征偏移匹配”图像偏移矫正技术和“边缘特征纹理动态自跟踪”的管径自跟踪监测技术,达到了既保证测量具有较高准确度,又提高了测量效率的目标。除上述检测技术外,本研究在数字图像处理的部分关键环节针对微循环图像的特征进行了有效改进。例如,(1)针对微血管与背景灰度特征的差异分析,提出了具有较高对比度的G通道首选方案,用于图像的分割处理。(2)针对单一传统图像分割技术在对背景复杂的肌肉组织来源的微血管图像进行血管提取时,易于连带提取组织背景,造成结果偏大的问题,提出多技术组合联用的"TPSP分割法”。(3)应用"TPSP分割法”,实现了微血管面积密度的自动测量,显著提高了测量效率。(4)基于人机交互方式,实现了微血管迂曲度和面积密度的计算机辅助自动检测,作为半自动化的检测手段,提高了测量速度。本研究主要创新点如下：1.本研究提出了一种专门针对肌肉组织来源图像的微血管网络进行自动分割的新组合方案—"TPSP分割法”,并在此基础上实现了微血管面积密度的自动化测量,显著提高了测量效率。该解决方案内容包括：以阂值法进行初次分割,以形态学滤波截断背景与微血管间连接区,以种子填充法对微血管区域进行二次识别与分割,再以形态学滤波填充分割区域中的孔洞和缺刻,实现了对肌肉组织来源图像复杂背景下的微血管网络的准确自动分割提取及血管面积密度测量。2.实现了部分微血管网络参数的人机交互式半自动检测。内容包括：计算机辅助自动读取标记点,自动长度测量、存储和换算,将原来完全手工测量迂曲度和面积密度的方式实现了半自动化,提高了测量速度。3.本研究提出了“位移模板自跟踪算法”,实现了对离体培养心肌细胞自主收缩运动的跟踪测量。能够同步检测心肌细胞的搏动频率、收缩幅度、收缩振幅、收缩速度和加速度多项收缩力度特征相关指标。4.建立了离体培养心肌细胞的收缩速度和收缩加速度定量指标,相对于搏动频率和收缩振幅该指标能够更好地表征收缩力学特性,与临床评价心脏收缩功能的方法更加贴近。5.应用本研究创建的心肌细胞搏动行为动态监测程序,对新生SD乳鼠心脏的心尖、心室和心房三个部位来源的离体培养心肌细胞的自主搏动进行自动化跟踪测量。结果表明：心尖来源的心肌细胞搏动频率显著高于心室和心房来源的心肌细胞,表现为高频低幅的收缩特性；而心室来源的心肌细胞的收缩振幅、收缩速度和收缩加速度显著高于心尖和心房部位来源的心肌细胞,表现为高幅中频的收缩特征,与其收缩做功的特性相符。这些差异是与心脏不同结构单元在心脏泵血功能的作用相适应的。6.本研究提出了一种消除微血管和微淋巴循环视频图像运动偏移的新技术方案—“区域特征偏移匹配”。该方案通过选定背景小区域局部特征模板作为图像偏移运动的跟踪目标,通过卷积相关算法连续动态提取图像序列各帧间的相对偏移矢量,以偏移矢量动态调整线形采样窗的位置,有效实现了对视频图像运动偏移所造成的采样位置偏差的矫正。7.本研究提出了微淋巴管、微血管自律运动边缘动态自跟踪及管径变化动态测量的新技术方案—“边缘特征纹理动态自跟踪”。在微循环视频图像运动偏移矫正的基础上，由人工指导确立首帧起始边界点和计算机程序自动生成边界特征模板,随后计算机程序自动进行边界特征相关匹配搜索,动态跟踪微淋巴管(或微血管)管壁边界位置的动态变化,有效克服了边界模糊给边缘识别及自动跟踪带来的困难,很好地实现了微血管和微淋巴管管径的动态自跟踪测量,极大地提高了微循环管径测量效率,为自律运动研究提供了有效的技术手段。更多还原

【Abstract】 Cardiovascular system is the motor system of maintaining basic vital movement. Autorhythmic systole/diastole changes (Vasomotion) are ubiquitous in the movements of heart, microvasculars and microlymphtics. Vasomotion is the energy source of tissue and organ perfusion. The autorhythm of circulation system is becoming one of the most hot spots of medical research.Cardiomyocyte is basic cardiac contractive functional unit on pumping blood. Dynamically analyzing the pulsation property of cultured cardiomyocyte possesses important significance on detecting the rule of autorhythmic movement and probing the pathogenesis of cardiovascular disease. The system of microcirculation which combined from microvasculars, microlymphtics and humour is critical part of circulatory system, and the main place of oxygen, nutrients, signaling molecules exchanging. Vasomotion of microvasculars and microlymphtics can not only adjust humoural stream but also afford energy to wavelike tissue perfusion.Vasomotion is a dynamic process versus time changes. A series of image sequence need to be analyzed to extract the characteristic parameters. High-speed, accurate automated dynamic measuring technology of characteristic parameters is the research foundation of biothythm. Because of limitations on the research techniques, some valuable indexes are difficult to extract, even can not be extracted at present. The objective of this research is proposing creative technical supports for autorhythmic activitiy studies.This study focused on measuring methods of characteristic parameters of static images and dynamic cardiomyocyte beating videos, autorhythmic microcirculation videos. Automated method was developed for extracting and measuring characteristic parameters of microvasculature. We also developed a dynamic, multi-parameter synchronous measuring method to study cardiomycyte autorhythmic activities in videos. This novel method has been used in the studies of mechanical parameters of cardiomycyte pulsation, and discovered some new phenomenon. In order to study on microcirculation vasomotion videos, we proposed a novel automated method to dynamically tracking and determining microlymphtic and microvascular diameters.Measurement of characteristic parameters in static images is the foundation of that in dynamic videos. At first, we analyzed the difference of gray degree between microvasculature and background of each color channel of colorful static microvascular images. The result shows that the channel G possessed the highest contrast, suggesting that channel G is the most adaptive arm for image processing. Single traditional algorithm of image segmentation cannot precisely extract vessel region from muscle tissue images with complex background. The result often bigger than true value, because of much background was misrecognized as vessel. We proposed a novel method called "TPSP segmentation algorithm" to overcome this obstacle. Channel G was choose to as processing arm. After noise filtering and image enhancing, image matrix was firstly segmented by threshold algorithm, morphological filtered to split vessel and remained background. Seed fill algorithm was employed to recognize the arm vessel region. Morphological filtered employed again to fill the lacunas caused by leukocytes and transparent plasma layer. And edges were smoothed by median filter. Finally, we realized the precise vasculature segmentation from muscle tissue images with complex background. Vascular density was calculated based on foregoing segmentation. Further more, we improved the burdensome manual measuring method of tortuosity and vascular density to be semi-automated, which can advance the speed of measurement.Autorhythmic cardiomyocytes are basic functional units. Because of the obstacle of lack of appropriate technology, studies of cardiomyocyte pulsatile activity mainly focus on frequency and intracellular electronic signal changes. In this study,"vector template auto-tracking algorithm " was employed to tracking the movement of arm point on cell surface. Not only can dynamically measure the frequency which can be determined by eyes, but also can measure mechanical parameters such as displacement, amplitude, velocity and acceleration which cannot be determined by eyes. Further more, we used this novel method to track the autorhythmic pulsation of48h cultured cardiomyocytes extracted from different part of neonatal SD rats. Result shows that there are significant statistic differences not only in constriction frequency but also in constriction amplitude, velocity and acceleration.Dynamic diameter changes of microvascular and microlymphtic is foundation of vasomotion research. Image sequence of microcirculation videos having the properties of image shifting and blur edges. In order to correct the error caused by image shifting, we employed feature tracking algorithm of moving objective to get the shift vector first. And then, adjust the position sample line according to the shift vector. We marked the initial edges and edge templates in the first image frame, and employed automated matching searches of edge features to track edge position changes dynamically. This a series of image processing makes we overcame obstacles in edge recognizing、edge tracking, and implemented the dynamic diameter tracking and measuring of microvasculars and microlymphtics.1. A novel image segmentation method-"TPSP segmentation algorithm" was proposed to extract vasculature from images which taken from muscle tissues. Vascular density was calculated based on segmentation. This method can significantly advance the speed of measurement. Image matrix was firstly segmented by threshold algorithm, morphological filtered to split vessel and remained background. Seed fill algorithm was employed to recognized the arm vessel region. Morphological filtered employed again to fill the lacunas caused by leukocytes and transparent plasma layer. And edges were smoothed by median filter. Finally, we realized the precise vasculature segmentation from muscle tissue images with complex background.2. We improved the burdensome manual measuring method of tortuosity and vascular density to be semi-automated, which can advance the speed of measurement.3."Vector Template Auto-tracking Algorithm" was employed to tracking the movement of arm point on cell surface. This method can measure mechanical parameters such as displacement, amplitude, velocity and acceleration which cannot be determined by eyes.4. Constriction velocity and acceleration were proposed as new indexs to evaluate constriction function of cultured cardiomyocytes. These indexs possess more coherence with clinical diagnosis on cardiac constriction function.5. We used the novel method to track the autorhythmic pulsation of48h cultured cardiomyocytes extracted from different part of neonatal SD rats. Result shows that there are significant statistic differences not only in constriction frequency but also in constriction amplitude, velocity and acceleration.6. In order to correct the error caused by image shifting, we employed feature tracking algorithm of moving objective to get the shift vector first. And then, adjust the position sample line according to the shift vector.7. We marked the initial edges and edge templates in the first image frame, and employed automated matching searches of edge features to track edge position changes dynamically. This a series of image processing makes we overcame obstacles in edge recognizing、edge tracking, and implemented the dynamic diameter tracking and measuring of microvasculars and microlymphtics.更多还原