Colon Cancer Classification of Histopathological Images Using Data Augmentation

Classification of colon cancer has great importance in medical diagnosis. An accurate and real-time investigation provides medical experts to review typical treatment timely. Histopathological inspection is a commonly used method to diagnose colon cancer. Presently, diagnosis methods depend on self-made features which take a long inspection period and require an expert medical professional. This paper proposes four convolutional neural network (CNN), namely, baseline CNN, two-block CNN, three-block CNN, and three-block CNN with data augmentation respectively, to classify colon tissue histopatho-logical images. The purpose of data augmentation is to check the efficacy of the proposed network compared to other existing methods. The histopathological images are given as input to four CNN. Three-block CNN with data augmentation achieved an accuracy of 99.40% shows that the proposed approach outper-forms other existing methods. The performance of the proposed network leads to an approach for precise cancer diagnosis.