【作者】 李晓宁；

【导师】 陈君楷；

【作者基本信息】 四川大学 ， 固体力学， 2003， 博士

【摘要】 通过各种力学装置对体外培养的活细胞施加机械力，是目前细胞力学研究的主要途径。在力学刺激下活细胞将发生多种适应性变化，不同的力学环境其变化的程度不尽相同。细胞形态学上的变化是最直观的变化，它预示并隐含着细胞功能上的变化。运用图象分析系统，定量研究细胞和亚细胞结构的运动和形态的动态变化特性，有助于理解力学刺激和细胞生长的关系，也有助于理解细胞病变的机理。但是，细胞力学实验中针对的是不能染色的生物活细胞图象，图象因细胞种类而异，并且图象性质随实验条件的变化而变化。迄今为止尚没有一个图象分析系统适用于所有的图象分析。鉴于此，本文针对交变应力环境下的细胞剪切力实验和细胞膜式张应力实验，开展了细胞力学实验图象处理平台和方法的研究。 针对剪切力环境下内皮细胞的形态变化，本文提出了两个反映细胞形态变化速率的指标，拉伸长度时间变化率和方向角时间变化率，可以综合表征细胞形态学变化的快慢。 图象分割是图象处理过程中最为困难也是最为关键的环节。本文深入分析了细胞力学实验中细胞图象的特点，针对细胞无交叠的图象，提出了基于微分边缘检测算子的图象分割技术路线，共有8个步骤。依次为图象增强、灰度化、消除光晕、Canny边缘检测、数学形态学膨胀、区域填充、平滑边缘、形态学重建。针对交叠现象严重的细胞图象，提出了基于watershed的的图象分割技术路线，共有7个步骤。依次为图象增强、灰度化、形态学光源校正分离细胞与背景、形态学重建去噪、距离转换、提取marke。图、water、hed转换，初步实现了对交叠细胞的分割。 本文使用ol帅pus IX70倒置相差显微镜观察细胞图象，采用两种采集方式采集细胞数字图象，建立了细胞力学实验的图象获取与处理硬件平台。一种采集方式是采用JvC一C68O一EC模拟CCO配合V工以g000V图象采集卡，另一种方式采用Pixera 1 50ES专业数字显微CCD。并且建立了显微镜照明自动控制系统，可以按照捕获设定，定时接通显微镜照明电源，减少显微镜照明灯泡的损耗。实现了细胞图象采集的程控化，使用者可以根据实验需要设定捕获方案，如单帧捕获、多帧捕获和视频图象流捕获，便于实现无人职守。 本文使用VisualC++ 6.0开发环境，建立了细胞图象处理的软件平台。它是一个集成化的软件包，将图像采集、文件管理、图像处理、特征提取、参数测量以及结果输出等多项功能有机地组织在一个集成环境中。由于C++集成了面向对象技术，该平台功能模块很容易实现功能扩充和代码移植。 运用建立的图象处理硬件平台，采集多个实验的细胞图象。结果表明，模拟CCD方案虽然分辨率不高，但是在采集细胞的视频图象方面具有优势:专用数字显微CCD方案分辨率高，光灵敏度高，图象采集数据损耗小，比较适合细胞静态图象采集。分别采用两种分割技术路线分割细胞图象，得到的细胞轮廓线与细胞实际边界比较吻合。运用开发的图象处理系统，开展了大鼠主动脉血管内皮细胞在病理脉动大剪切力环境下的形态学研究，实验结果表明，图象处理系统可以胜任细胞的形态学参数的定量分析。更多还原

【Abstract】 Because of the staggering complexity of the in vivo environment, systematic study of the cellular response to mechanical stimulations has relied heavily on the use of in vitro preparations. Living cells under mechanical stimulations can make adaptive changes and it must be recognized that different biomechanical stimulations can induce different cellular changes. Morphological change is the most instinctive one and it can also be an indicator for the functional changes in cells. The quantitative study on cellular and sub-cellular morphology and movement by image analyzing system is useful not only in understanding the relations between force simulation and cellular growth, but also in catching on cellular pathological mechanism. However, few image analysis systems are found to suitable for all kinds of images, since images, especially the biological living cellular images, which are sourced by different methods have different characteristics. According to this condition, a new image-processing platform to analyze the cellular changes induced by shear stress and tensile stress has been set up in our lab and the image-processing methods for cellular biomechanical experiments have been detailed studied in this dissertation.The morphological parameters of endothelial cells exposed on shear stress are analyzed in detail and then two novel parameters are suggested. They are named VOA (velocity of oriental angle) and VL (velocity of major axis length) and can reflect velocity in cellular morphological changes.Image segmentation plays a key role in image analysis and it is also the most hardly step in image processing pipeline. As for all the non-overlapping images, one method which has eight steps based on difference arithmetic operators is proposed. Image enhancing, image graying, halation removing, Canny edge detection, mathematical morphology dilation, edge smoothing, mathematical morphological reconstruction are implemented according to the set sequence. As for all the seriouslyoverlapping images, another method which has seven steps based on watershed transform is put forward. These steps in turn are image enhancing, image graying, morphological illumination, mathematical morphological reconstruction, distance transform, marker image extraction, watershed transform.Olympus 1X70 phase contrast inverted microscope is used to observe cellular images and two methods are applied to capture digital images in the construction of image processing hardware. One is through video CCD (JVOC680-EC) equipped with image capture board (VIGA 9000V), the other is through professional digital CCD for microscopy (Pixera 150ES). A system for automatic control illumination of microscope is set up, which can put through the illumination power of microscope according to the set capture schedule and will help decrease the loss of the microscope’s bulb. And the images can be captured in a manner of single frame, frame-by-frame or video stream.The cellular mechanical experimental image-processing platform is developed by using Visual C++ 6.0 (Microsoft) programming language. It is an integrated software package, which integrates image acquiring, files managing, image processing, feature extracting, parameter measuring and result outputting together. Because Oriental Object Programming (OOP) skills are applied to this system, the modules in this package are easy to accomplish functional expanding and code implanting.Enormous cellular images are captured through this image-processing platform and the results suggest that video CCD schedule has advantages in capturing video stream with the disadvantages in the image resolution, and digital CCD is very useful in capturing still cellular image because of its relatively high resolution, high sensitivity and less data loss. The superposition of the processed image by either segmentation strategy on the original cell image indicates that the two methods are suitable. By using developed image analysis system, the morphological parameters of rat aortic endothelial cells (RAECs) expo更多还原