基于顺序形态变换的生物芯片图像网格化  

Microarray gridding based on order morphology transform

作　　者：付东翔[1] 马军山[1] 陈家璧[1] 庄松林[1] 

机构地区：[1]上海理工大学光学与电子信息学院

出　　处：《光学技术》2006年第6期912-915,919,共5页Optical Technique

基　　金：上海市教委重点资助项目(05ZZ26);上海市教委青年基金资助项目(05EZ46);上海市重点学科建设资助项目(T0501)

摘　　要：生物芯片图像网格化是对由生物芯片扫描得到的荧光成像中的信号（Foreground）和背景（Background）进行划分，以便提取各点的吸光度值、面积和吸光度比等荧光点的数据。网格化的准确与否决定了数据提取的正确性。提出了一种运用顺序形态变换理论实现生物芯片荧光图像网格化的方法，通过对万分位值在0～1之间的顺序形态进行变换，可实现像素灰度的极小值、中间值等多刻度值的灰度取值运算。理论分析表明，该算法能够识别信号与背景的边界，可去除噪声等非边界信息，能判断信号区域，可确定荧光点的边界和范围，可实现荧光点阵图像的网格化。试验结果表明，该算法对不同信噪比的荧光图像来说有很强的适应性，可用于高密度生物芯片图像的处理。Gridding of fluorescence image of microarray is to identify foreground and background of fluorescence spots. Ab sorbency, area and absorbeney ratio of each fluorescence spot can be operated after gridding. So it is important that result of gridding is accurate. A algorithm is proposed to process and grid microarray by order morphology transform. Multl-scale pixel gray operatiorns are done with different percent which belong to [ 0, 1 ]. It is proved that scope and boundary of fluorescence spots can be found by analyzing profile of gray, ale. Then vertical and horizontal grids are located. Data analysis and experiments show that the way is adaptive and robust to different SNR fluorescence image of microarray, especially high density micoarray.

关 键 词：顺序形态学 生物芯片 网格化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]