基于数字图像处理的基因芯片信号点检测和分析系统

【作者】 袁湘；

【导师】 杨欣荣； 罗桂娥；

【作者基本信息】 中南大学 ， 计算机应用技术， 2005， 硕士

【摘要】 芯片分析是基因芯片技术的一个重要组成部分,在基因芯片的应用中扮演着一个重要的角色。在本论文中,主要研究基因芯片图像的分析和数据处理,从而提取出基因芯片的杂交反应信号点的亮度信息,获得有用的生物信息。 本文介绍了基因芯片技术在国内外的发展现状和应用前景,提出现阶段基因芯片技术在应用上所遇到的问题,从而提出了课题的研究背景和课题来源、意义。然后结合基因芯片数字图像的特点,给出了图像信号点分析的基本步骤。 由于基因芯片在杂交过程中会产生一些噪声,如某些无规律的噪声、载玻片上的灰尘等干扰,将会影响到信号点亮度的精确提取。本文探讨了几种常用的滤波方法,如平滑滤波、顺序滤波等,并选取一种自适应中值滤波的方法去除掉图像背景中那些无规律的噪声干扰。 为了确定每一个信号点的区域,必须对图像的每个信号点进行栅格化处理。通过图像水平投影和垂直投影的方式确定了每个信号点的位置区域,使得在每一个固定大小的区域之内只包含唯一的一个信号点。 在获得信号点亮度区域的方法中,本文重点讨论了信号点边缘检测和边缘连接的方法。根据图像灰度变化的趋势,运用梯度算子模板对图像进行检测,可以获得信号点的粗边缘信息;然后运用拉普拉斯算子模板实现信号点边缘的精确定位;最后根据一定的连接准则,将边缘点连接成一个完整封闭的信号点边界。 在信号点的亮度信息提取中,主要讨论亮度的提取方法,背景信息的移除,数据的归一化处理等。本文采用一种种子扫描线区域填充的方法提取信号点的亮度信息,然后通过背景移除和数据的归一化处理,得到了信号点的亮度值。 根据以上的原理,结合芯片图像的特点,开发了一个基因芯片数字荧光图像检测和分析系统。可以完成图像滤波,快速栅格化,边缘检测和亮度信息提取等功能,最终实现图像信息的快速、准确的提取。更多还原

【Abstract】 Chip analysis is the key part of techniques for Gene-chip and plays an important role in the application of Gene-chip. In this paper, we study mainly the methods of Gene image analysis and data processing. It can extract the hybridizing intensity and obtain the useful biological information.At first of the paper, we introduce the development of Gene-chip and the prospect of its application at home and abroad. And then we raise problems existing in the application of Gene-chip at present. So we put forward the research background, source and significance of the task. Combining with the traits of Gene-chip image, we provide the basic steps to analysis signal points in a image.Firstly, due to some noise from the hybridizing operation, such as irregular spots, dust on the slide, it can make it difficult to extract data accurately. We discuss several general filtering methods, such as smooth filtering, median filtering. Finally, we choose a self-adaptive median filtering which can eliminate successfully the irregular noise from background and remain the details of image.Secondly, to acquire the grid region of a signal point, we propose a grid processing for every signal point in the image. By horizontal projection and vertical projection of the signal points in the image, we can locate the grid region of every signal point so that there is only one signal point in a grid region.We discuss deeply the methods of edge detection and edge linkage for obtaining the target region of a signal point. Based on transition on gray level of the image, we can define a pixel point as an edge point by using its two-dimension first derivative, and define accurately the central position of an edge by using its two-dimension second derivative. Finally, according with some special standards, we can connect the right points which accords with the standards to form an edge.In extraction of fluorescent intensity, we discuss mainly the methods of intensity extracting, background removing, date computing and normalization. We can use a seeded region-fill by scanning line to extract the intensity of the signal point. After removing background and date normalization, we can get the intensity information of the signal point.Finally, we develop a software system for detecting and analyzing the signal points in Gene-chip image. It can provide image processing, image filtering, fast gridding, edge detection, intensity extraction and so on. Lastly, we can obtain rapidlyand accurately the biological information from Gene-chip image.更多还原