Fold contour estimation technique for pre-processing 3D colonoscopy images: Polyp cases

Abstract

Colon fold mapping is useful as the basis of a 3D reconstruction from 2D colonoscopy single image. The existence of abnormality, such as polyp, will cause errors, since this method is based on edge detection. In this paper, a novel algorithm to determine the colon folds of colonoscopy images is introduced. The proposed methods divided into three main steps. The first step is the preprocessing with removing of specular highlight, blood vessel, and shading. The second step is selecting the actual colon fold, which is done by identifying the fold characteristic using curvature analysis, and compared it with the polyp characteristic. The final step is closing the colon fold contour. The preliminary results show that the proposed algorithm forming a colon fold correctly, although there are polyps in the intestines. Some consideration remains for further development of colon polyps fold estimation in case.