【摘要】 本文描述一种自动分割双重标记免疫组化图像的方法,此方法直接从双重标记的两种阳性组织细胞着色特征对其进行分离。与常用的彩色反卷积方法相比,此方法提高了分割准确度且需要样品切片的数量从3张减少到1张。通过细胞角蛋白和S-100蛋白双重标记,分别经二氨基联苯胺(DAB)与3-氨基-9-乙基咔唑(AEC)显色的舌鳞癌样品图像的实验证明,此方法可以有效地将鳞癌与神经组织细胞两种不同的染色区域分割出来,与手动分割的结果作比较其准确度和精确度都有较大的提高。此方法也适用于分割其他用DAB和AEC双重染色的免疫组化图像。更多还原

【Abstract】 This paper describes a method for automated segmentation of double stained immunohistochemical tissue images that overcomes the limitations of the manual approach as well as of the existing techniques.The method is based on color separation and morphological processing.Compared with the commonly used color deconvolution method,this method improves the segmentation accuracy and only needs to make one section sample instead of three which are necessary for color deconvolution.Experimental results on the tissue images of the squamous cell carcinoma of tongue labeled with cytokeratin(CK) and S-100 protein,stained with 3,3′-Diaminobenzidine(DAB) and 3-Amino-9-Ethylcarbazole(AEC) show that this method can effectively segment cancer cells and nervous cells.In addition,this experiment demonstrates high degree of accuracy and precision compared with manual segmentation.更多还原