基于色度学准则分析的免疫组化彩色图像C-均值聚类分割技术研究

A technical study on C-mean clustering segmentation for color immunohistochemical image based on matlab

免疫组化彩色图像分割在免疫组化染色定量分析中有重要的应用价值。C-均值聚类算法(CMA)是根据一定的相似性准则将图像分成C类。由于运算十分耗时,直接限制了C-均值聚类算法在彩色图像分割领域的应用。本文针对免疫组化彩色图像特点,提出了分割免疫组化彩色图像的色度学准则,即:用每个像素的R分量减去B分量,根据其差值是否大于0将相应的像素分为两大类:(R-B)≥0类和(R-B)<0类,自动分离出图像的阳性细胞区域和阴性细胞区域。在此基础上我们改进了C-均值聚类分割的方法:①针对上述两大类图像的色彩分布的特点确定初始聚类中心;②分别对上述两大类像素在一个色度学空间聚类;③根据每次迭代过程中聚类中心的变化趋势来预测下一次迭代后可能的聚类中心,从而减少迭代的次数。改进之后的C-均值聚类分割方法减少了聚类的样本数量,降低了算法的复杂度,且由于迭代次数的减少,运行速度得到了提高。实验结果表明,本文建立的技术方法能有效地分割免疫组化彩色图像。

Immunohistochemical image segmentations play an important role in the immunohistochemical staining quantitative analysis. C-Means clustering algorithm （CMA） is a method to partition of an image into homogeneous regions. According to characteristics of color immunohistochemical images, a chroma criterion and improvement of CMA for the segmentation of immunohistochemical image was proposed to solve the problems of heavy calculating burden if applying CMA directly to real color immunohistochemical image segmentation in three color spaces. The chroma criterion is that to calculate the subtraction of R value and B value for each pixel first, and then divide the pixels into two large classes. Thus, positive cell area and negative cell area are separated automatically. On the basis of the results, the CMA algorithm is improved in two aspects： （1） The original class center is obtained by color arrangement feature; （2） the CMA is executed apart on two classes pixels in one color space; （3） The next iterative center can be conjectured on change trend after each iteration, which reduces the times of iteration. The improvements lessen sample amounts, reduce algorithm complexity, decrease iteration times and speed up calculation. The results reveal that the technique is effective.