Automated classification of local patches in colon histopathology

An automated histology analysis is proposed for classification of local image patches of colon histopathology images into four principle classes: normal, cancer, adenomatous and inflamed classes. Shape features based on stroma, lumen and imperfectly segmented nuclei are combined with texture features for classification. The classification is analyzed under the three scenarios: normal vs. abnormal, cancer vs. non-cancer and four-class classification on a labeled dataset consisting of 2000 patches per class which were collected from 55 different slices. The proposed method achieves 79.28% mean accuracy between normal and abnormal; 87.67% accuracy between cancer and non-cancer and 75.15% between the four classes with equal class priories.