【作者】 谢杰镇；

【导师】 陈朝； 王博亮；

【作者基本信息】 厦门大学 ， 凝聚态物理， 2007， 博士

【摘要】 生物体基本上是由软凝聚态物质组成,生物膜是软凝聚态物质物理研究的重要内容。本论文研究选定人眼角膜――一种典型的生物膜为研究对象展开。角膜是人眼的重要组成部分,角膜的特性对于眼疾病的诊断和治疗有着重要的意义,研究角膜的形态特征、物理特性,对其进行建模与仿真具有重要的学术意义和临床应用价值。本文主要研究了在构建中国人虚拟眼角膜的形态和物理模型过程中所遇到的图像采集、分析、参数提取、形态建模、物理建模和虚拟仿真等方面的问题,探索一条将数字人眼组织模型数据与临床个体数据融合的研究路线,其目的是构造出能准确描述角膜的形态结构和物理模型,为构建“中国人虚拟眼”以及眼科医学教学和临床应用奠定重要的物质和技术基础。本论文的主要研究内容和创新性工作如下:1.负责设计并实现了一套从普通眼科临床使用的裂隙灯显微镜采集和分析眼前节显微图像的软硬件系统,提出了一种基于裂隙灯图像客观地获取眼前节的10多种生理参数和全自动分析报告眼角膜主要生理参数的新方法,设计了一种基于临床裂隙灯显微序列图像的角膜三维重建的方法。利用Directshow技术采集裂隙灯显微图像,采用等效的大津法阈值分割出角膜和虹膜区域,然后根据角膜和虹膜组织结构的边界拟合曲线提取眼前节的特征参数,获取并分析报告临床诊疗相关的眼角膜、虹膜、晶状体等组织主要生理参数。并以裂隙灯图像序列为基础,利用表面绘制方法实现基于病人裂隙灯图像的角膜三维重建。2.原创性地制作并保存了646片眼组织切片标本,以此为基础获得了眼球切片一套五种图像基础数据集和切片角膜组织的图像数据集。作为团队的最主要骨干,负责实现了原创性地将一只眼球进行包埋、组织切削、染色并制成646片组织切片标本,以此为基础获得了眼球切片一套五种图像基础数据集和切片角膜组织的图像数据集,这是至今为止国内外一只眼球制成并保存组织标本数量最多、数据量最大的成果。3.将经典的角膜数据和由裂隙灯图像所获取的临床人眼角膜数据进行融合,提出一种建立参数化的角膜形态模型和有限元物理模型的方法。结合经典的角膜数据和由裂隙灯图像所获取的临床人眼角膜数据,在合理简化和假设基础上采用参数化的建模方法,建立参数化的角膜形态模型和有限元物理模型,并进行了眼内压和外部压力改变时的角膜形变仿真研究,探索出将数字人技术为临床病人服务的研究路线。4.建立人眼角膜的弹簧-质点模型,结合力反馈设备,对人眼角膜组织进行了接触虚拟仿真研究。针对眼角膜的基本几何和物理特征,设计了以AABB层次包围盒树结构为碰撞检测的基本数据结构,并改进了弹簧-质点模型,实现了虚拟眼角膜接触变形和触觉反馈仿真。实验表明,所提出的算法效率较高,在视觉和触觉上实现实时的绘制和反馈。很好的表现了人眼角膜的基本特征,为进一步建立虚拟手术仿真环境奠定了基础。更多还原

【Abstract】 Soft matter physics such as bio-membranes play an important creative role in basic science in the coming century. Organisms are basically composed by the soft material. Virtual organ is a very important research domain of modern biomedical engineering. The cornea is a transparent external surface, which covers the front of the eye. It is a powerful refracting matter, providing 2/3 of the eye’s focusing power. So, the construction of virtual eye cornea has very important theory meaning and application value for the ophthalmological physiology, pathology, and clinic.In this paper, in order to construct form model and physical model of virtual eye cornea, the problems of image capturing, segment, analyzing, physical modeling and simulation which are involved in were discussed. And a research route that fuses the clinical data and digital human technology into the model of human eye was explored. This work is an important component of virtual Chinese human eye research.The main contents and contributions of this dissertation are as follows:a. A system which can be used to capture and analyze slit light micrograph from the ophthalmological slit lamp microscope was built. A new automatic method was proposed to analyze the micrograph of anterior eye, and some proper results were got. A method of 3D reconstruction of cornea from slit lamp images was proposed. Slit lamp microscope images were capturing with the DirectShow technology. Edges of the outside, intestine surface of the corneal and the anterior surface of the iris in the image were segmented. Some biologic parameters of anterior segment such as curvature of cornea surface, thickness of Corneal, and deepness of anterior chamber, were given automatically by curve fitting. The system can be provided as an important diagnostic tool for the doctor in the clinic. It is convenient and practicable with good intercommunion.b. One eyeball was solidified, embedded in collodion, sliced and dyed by the H.E method. 646 pieces of slice were got first time. Five datasets were established base on eye tissue slices. And images of cornea were acquired. This is the largest number of slices and the largest data of eye tissue by now. Main work was done in this process.c. A method of building a physical model of the cornea with finite element method was proposed. An individual 3D modal model of cornea was constructed. Some parameters of model were got by the slit light micrograph system. The stroma is the thickest layer and is the main load bearing parts. So the material parameter of stroma was considered in the paper. The finite element model was built up based on finite element theory. The intra-ocular pressure and external pressure were simulated, and the cornea shape changes computed using finite element analysis.d. A deformation model of virtual cornea based on mass-spring model was presented. A collision detection and response method which is adapted to surface models of cornea was proposed. The model is based on half-edge data structure and hierarchies of axis-aligned bounding boxes (AABBs) to satisfy the higher frame rates in haptic feedback. The method is simple and effective whether in theory or in practice. The experiment results show that the presented method is faster than AABBs method used solely. It creates the foundation for a virtual surgery simulation environment.更多还原