Effective segmentation and classification for HCC biopsy images
Effective segmentation and classification for HCC biopsy images
复制标题
DOI：
10.1016/j.patcog.2009.10.014
复制
发表时间：
2010-04-01
影响因子：
8
通讯作者：
Lai, Yan-Hao
中科院分区：
文献类型：
--
作者：
Huang, Po-Whei;Lai, Yan-Hao
肝细胞癌活检图像的准确分级对预后和治疗计划具有重要意义。在本文中，我们提出了一个自动分级的肝癌活检图像系统。在预处理过程中，我们采用了双重形态灰度重建的方法来去除噪声，突出细胞核形状。利用标记控制的分水岭变换得到细胞核的初始轮廓，利用Snake模型平滑、准确地分割出细胞核的形状。然后根据六种类型的特征提取14个特征用于肝癌分类。最后，我们提出了一种基于支持向量机的决策图分类器对肝细胞癌活检图像进行分类。实验结果表明，基于支持向量机的决策图分类器的分类正确率为94.54%，而k-NN和支持向量机的分类正确率分别为90.07%和92.88%。(C)2009爱思唯尔有限公司。保留所有权利。
Accurate grading for hepatocellular carcinoma (HCC) biopsy images is important to prognosis and treatment planning. In this paper, we propose an automatic system for grading HCC biopsy images. In preprocessing, we use a dual morphological grayscale reconstruction method to remove noise and accentuate nuclear shapes. A marker-controlled watershed transform is applied to obtain the initial contours of nuclei and a snake model is used to segment the shapes of nuclei smoothly and precisely. Fourteen features are then extracted based on six types of characteristics for HCC classification. Finally, we propose a SVM-based decision-graph classifier to classify HCC biopsy images. Experimental results show that 94.54% of classification accuracy can be achieved by using our SVM-based decision-graph classifier while 90.07% and 92.88% of classification accuracy can be achieved by using k-NN and SVM classifiers, respectively. (C) 2009 Elsevier Ltd. All rights reserved.