Detection of Collagenous Colitis Based on Histopathology Image Segmentation of Colon

Collagenous colitis is prevalent among pathological diseases. Because of normal radiological and endoscopic appearances, the microscopic image of colon tissue is necessary and important material for investigation of this disease. In this paper, a computerized system is developed for collagenous colitis diagnosis. First, colon tissues are colored with Mason trichrome to make sub-epithelial collagen band to be appeared in blue. Second, preprocessing is done on the microscopic image of colon tissue, and then Color histogram features in HSV color space and Haar Wavelet energy in detail coefficients of sub-band images are implemented to characterize the histological structure representation for tissue classification. KNN Classifier is selected for this classification, which segments the sub-epithelial collagen band from the image, and finally, thickness of sub-epithelial collagen band is determined by morphological operations. Abnormal case will be recognized if the thickness of sub-epithelial collagen band be more than 10 μm. Fifty clinical image samples were employed in the training and testing procedures for the evaluation of system performance. Through the applied classification it was revealed that collagenous colitis can be diagnosed accurately.