Cancer Screening On Indian Colon Biopsy Images Using Texture and Morphological Features

Colon Cancer Detection in the biopsy images is the tedious task for the pathologist. Thus an automation in this regard may speed up the processs and can assist the pathologist. This work focus on the classification of the colon biopsy images to normal and cancerous ones where the hybrid of the texture and morphological features are taken from the Indian Population at different magnification factors. The hybrid of the texture features are taken from the Harallick features, HOG, Histogram, GLRLM, LBP and Gabor. The classification was carried out with classifiers SVM, Nave Bayes, Adaboost, Multilayer Perceptron and Decision Tree. The system has been tried with two datasets which was taken from Aster Hospital, Kochi and HCG Hospital, Bengaluru where the fused feature set of hybrid texture and morphological feature gave a good performance with the classifiers SVM and Multilayer Perceptron.