CAM血管新生图像处理算法研究

发布时间：2018-01-06 16:09

本文关键词：CAM血管新生图像处理算法研究　出处：《长春工业大学》2016年硕士论文　论文类型：学位论文

【摘要】：血管新生(angiogenesis)就是从现有血管中生长出新血管的过程。血管新生不仅在正常生理变化中(如生长、伤口愈合)起着重要的作用,而且和癌症的机理也有着紧密关系。因此,研究血管新生对癌症的治疗和预防具有重要意义。血管生长水平一般通过计算一定区域的血管长度、血管面积、分叉点数等量化参数来评估。实验材料易得、实验周期短、操作简便并且结果易于观察的鸡胚绒毛尿囊膜(CAM)是研究血管新生活动的常用实验载体。针对CAM血管图像具有目标和背景区域对比度低、局部血管出现断裂、不同血管区域的灰度值差距较大等特点,通过Otsu阈值分割算法结合OPTA细化算法获得初始的种子点,然后通过迭代阈值跟踪算法进行图像分割,提取血管信息。与传统的单一阈值分割法、边缘检测分割法相比,本文迭代阈值跟踪算法能较好的分割出CAM血管图像中细小血管信息,并且检测到的血管信息更加准确和完整。骨架信息提取的好坏直接影响后面量化评估。针对分割后的CAM血管图像具有宽度不一、分支点多等特点,通过Hessian矩阵确定血管方向和离散高斯核函数卷积CAM血管图像获得血管骨架点信息。对比发现,本文骨架提取算法明显优于传统细化算法,并且具有良好的中轴性和连通性。为了更加直观地看到每个处理模块的效果,最后设计了GUI系统。对比实验结果可以看出本文量化结果和专家手动获得的量化结果误差很小,并且可重复性应用。
[Abstract]:Neovascularization (angiogenesis) is a process of growth of new blood vessels from existing blood vessels. Angiogenesis not only in normal physiological changes (such as growth, wound healing) plays an important role, and the mechanism of cancer and also has a close relationship. Therefore, it is important to study angiogenesis treatment and prevention of cancer. The general level of blood vessel growth by calculating the area of vascular length, vessel area, number of quantitative parameters to assess the bifurcation. The experimental materials are easy to obtain, the experimental period is short, simple operation and easy results chorio urine capsule observation (CAM) is a commonly used experimental carrier to study angiogenesis activity. For CAM imaging with goal and the background of low contrast, local vascular rupture, gray vascular region is a big gap between the different features, through the Otsu threshold segmentation algorithm based on OPTA algorithm to obtain the initial refinement The seed point, and then through the iterative threshold tracking algorithm for image segmentation, extraction of vascular information. And the traditional single threshold segmentation method, edge detection segmentation method, the iterative threshold tracking algorithm can better segment the small blood vessels CAM angiography image information, and detected the vascular information more accurate and complete information extraction framework. The direct impact behind the quantitative evaluation. For CAM vascular image segmentation after having a width of a branch point features, using Hessian matrix to determine convolution CAM imaging vascular direction and discrete Gauss kernel function to obtain vascular skeleton point information. By comparison, the algorithm is obviously superior to the traditional thinning algorithm to extract the skeleton, and good the axis and connectivity. In order to see the effect of each processing module is more intuitive, the final design of the GUI system. The experimental results can be seen The quantitative results obtained by this paper and the manual obtained by the expert are very small, and can be used repeatable.

【学位授予单位】：长春工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R730.4;TP391.41

【相似文献】