医学显微图像配准及存储技术研究

发布时间：2018-02-26 08:32

本文关键词： 图像配准图像拼接显微图像 BigTIFF 虚拟显微镜　出处：《华中科技大学》2013年博士论文　论文类型：学位论文

【摘要】：多年来,病理专家通过传统光学显微镜观察病理组织的玻璃切片来进行诊断和研究。如今,病理专家可以在计算机上通过虚拟显微镜软件来观察数字切片,诊断和研究的过程更为便捷、高效。生成数字切片需要用自动显微镜扫描玻璃切片然后拼接扫描所获取的显微图像序列,其核心技术包括显微图像的采集、配准、融合、拼图的存储和显示等。本论文针对其中的显微图像配准、拼图的存储和显示这两个核心问题展开研究。论文的主要工作包括：第一,针对扫描获取的大量(数百乃至数千张)显微图像中相邻图像的快速配准问题,提出一种对数差值点集模板方法。论文定义像素灰度值对数差值函数来描述显微图像像素间的灰度变化,基于该函数从显微图像中挑选特征点构成模板,然后通过模板匹配来配准相邻显微图像。该方法配准速度快,且具有自动检查配准结果的功能。第二,针对眼球等含有大片空白区域的病理切片的显微图像配准问题,提出一种基于图像信息量的显微图像序列配准路径规划方法。该方法先通过颜色信息熵快速计算各显微图像的信息量,然后按图像信息量大小规划配准路径。与传统的序列配准方法相比,该方法能防止空白显微图像或空白重叠区域参与配准而导致的配准错误,从而使序列配准正确、高效。第三,针对大量显微图像拼接出的上GB的超大图像存储、显示问题,以BigTIFF图像文件格式为基础,提出一种专用于数字切片的图像文件格式Medical BigTIFF (MBTF)。其优点是可以存储的图像数据量仅受计算机存储设备限制,可存储相关医疗信息,并且通过其层次存储和访问模型可以按观察点快速加载数字切片并显示。最后,利用本论文研究的技术开发了一套病理切片数字化扫描原型系统,并对四周存在空白区域的病理切片、内部包含大片空白的病理切片和稀疏分布的病理涂片等不同特点的病理切片进行了数字切片扫描生成实验和数字切片显示实验,实验结果显示了本论文提出的方法的优点。
[Abstract]:Over the years, pathologists have been able to diagnose and study glass sections of pathological tissue through traditional optical microscopy. Today, pathologists can use virtual microscope software to observe digital slices on a computer. The process of diagnosis and research is more convenient and efficient. To generate digital slice, we need to use automatic microscope to scan glass slice and then splice the sequence of microscopic image. The core technology includes the acquisition, registration and fusion of microscopic image. The storage and display of jigsaw puzzle are studied in this paper. The main work of this thesis is as follows:. First, for the problem of fast registration of adjacent images in a large number of microscopic images (hundreds or even thousands) obtained by scanning, A method of logarithmic difference point set template is proposed in this paper, the gray value logarithmic difference function of pixel is defined to describe the change of gray level between pixels of microscopic image, and based on this function, feature points are selected from microscopic image to form template. Then the adjacent microscopic images are registered by template matching, which is fast and has the function of automatically checking the registration results. Second, aiming at the problem of registration of microscopic images of pathological sections with large blank areas, such as the eyeball, A path planning method for registration of microscopic image sequences based on the amount of image information is proposed, in which the amount of information of each microscopic image is calculated quickly by using the color information entropy. Then the registration path is planned according to the amount of image information. Compared with the traditional sequence registration method, this method can prevent the registration errors caused by the participation of blank microscopic images or blank overlapped regions, so that the sequence registration is correct and efficient. Third, aiming at the storage and display problem of large GB images on large GB images stitched out by a large number of microscopic images, the paper bases on the BigTIFF image file format. A special image file format for digital slicing, Medical BigTIFF / MBTFN, is proposed. The advantage of this format is that the amount of image data can be stored only by the computer storage device, and the relevant medical information can be stored. Through its hierarchical storage and access model, the digital slice can be quickly loaded and displayed according to the observation point. Finally, a set of digital scanning prototype system of pathological section is developed by using the technology studied in this paper. The experiments of digital section scanning and digital section display are carried out for pathological sections with different characteristics, such as large blank sections and sparse distributed pathological smears. The experimental results show the advantages of the proposed method in this paper.
【学位授予单位】：华中科技大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：R318.6;TP391.41

【相似文献】