Automatic Detection of Histological Components in Breast Cancer Image by Using Genetic Neural Network

摘要

We show a new approach for effective segmentation method of pathology components in digitized tissue image to support computer-aided histological inspection. Because of the requirement of high accuracy in the area for precise diagnosis of disease, an adaptive and automated segmentation scheme of different component in stained cancer cells has been investigated. We employ a new approach for adaptation to many kind of components by using neural network with genetic optimization process. The proposed method adopts genetic vector quantization based on RCE network for realizing appropriate segmentation in microscopic image. Because our former presented method has no ability to adjust its structure, in this paper, we employ genetic adaptation scheme. Experimental results of extraction of specific components from cancer cell image and its performance are described.