Cancer detection from microscopic biopsy images using image processing and pattern recognition tools: A review

Abstract

Presently cancer detection is done by pathologists through evaluation of microscopic biopsy of cancerous tissues by examining the tissue structures, distributions of cells in tissue, regularities of cell shape to determine the level of abnormalities that may be present in the sample under investigation. The outcomes of these examinations may be normal, benign and malignant tissues. The manual evaluation of microscopic biopsy for cancer detection leads to subjective, time consuming and varies with perceptions and level of expertise of pathologists. To overcome these challenges automated cancer diagnosis is needed for objective, fast, accurate and quantitative results. In this paper, a systematic survey on computational steps for detection of cancer from biopsy images using image processing and pattern recognition tools is presented. These steps involve image preprocessing, enhancement and restoration, segmentation, feature extraction to quantify properties of local area, and classification of sample image into normal and abnormal categories e.g., benign and malignant ones. Copyright © 2015 American Scientific Publishers All rights reserved.