WSEAS Transactions on Biology and Biomedicine
Print ISSN: 1109-9518, E-ISSN: 2224-2902
Volume 11, 2014
A Computer-Aided System for Discriminating Normal from Cancerous Regions in IHC Liver Cancer Tissue Images Using K-means Clustering
Authors: , , , , , , ,
Abstract: Immunohistochemistry (IHC) is a well established imaging technique that can be exploited to detect whether the target antigen exists in tissue sections or not in order to discriminate between the cancerous and normal regions in a cancer tissue specimen. The intensity of immuno-stained protein in normal and cancerous regions can be compared to detect the gene status in sample tissues. In this paper, we address the problem of identifying the differential expression of marker protein on cancerous and normal regions in an IHC liver cancer tissue image. We present an improved IHC image processing procedure based on nucleus density, intensity of stained protein, and k-means clustering algorithm to develop an automated system for analyzing an IHC image. The proposed system can discriminate between normal and cancerous regions in an IHC image more effectively and display them visually. Furthermore, this system can automatically evaluate the stained protein expressions in the two regions which can help the pathologist to analyze the differential expressions of a marker protein from IHC images. Finally, we evaluated the proposed system on 150 real IHC liver cancer tissue images and compared the results with those obtained using support vector machine (SVM) and a previous work where the average of density of nuclei is used as the threshold to discriminate cancerous from normal regions.
Search Articles
Keywords: Immunohistochemistry, Tissue Image, Nuclei Segmentation, K-means Clustering, Support Vector Machine
Pages: 29-34
WSEAS Transactions on Biology and Biomedicine, ISSN / E-ISSN: 1109-9518 / 2224-2902, Volume 11, 2014, Art. #5