Classification of Colon Polyp on Endoscopic Image Using Support Vector Machine

Colon polyps are abnormal tissues belonging to tumors that grow on the walls of the human colon. Colon polyps can be visualized using the technology that is endoscopy. However, the results of endoscopy image examination often differ depending on the accuracy and experience of the doctor. To overcome these problems, a system was built by implementing digital image processing that can detect types of colon polyps based on endoscopy images. There are several stages of the process in this system. The first process is preprocessing using scaling, and bilateral filtering which is used to improve endoscopic image quality. Then, the the second process is segmentation to separate the area of polyp from other object using watershed, thresholding, and cropping. The third process is extraction used to get the area, length, width, and texture of the polyp. The last process is classifying colon polyps based on their type is a hyperplastic, adenoma, and sessile using the SVM (Support Vector Machine) method. The system in this study was tested on 20 endoscopic images, and this system has a high level of accuracy, sensitivity, precision, and specificity in classifying colon polyps. is 93.54%, 94.73%, 85.71%, and 93.02% respectively.