Maturity Differentiation of Colon Polyp Based on Endoscopic Images Using Machine Learning and Spatial Feature

Colon cancer is a dangerous type of degradative disease after lung cancer. Colon cancer start with the presence of polyps on the colon. Early preventive measures are important anticipatory steps to prevent the development of polyps into cancer, as well as being an important point in determining the treatment that must be undertaken by patients. Early detection of polyps based on medical images is very challenging because it involves instance factors and the technical factors used. This study aims to produce an automation system classification of polyp maturation so that it can facilitate pathologists and health practitioners in diagnosing polyps in the human intestine and determine the level of maturation early. This study based on a colonic endoscopic dataset utilizing spatial feature extraction to extract important features from the polyp image and feedforward neural network backpropagation classifier function with a hidden layer to determine its maturation class. Based on experiments conducted, it found that the proposed system was able to classify polyp maturation classes with an accuracy level of 93.94% and a false-positive rate (FPR) level of 6.05%.