AUTOMATIC POLYP DETECTION FROM CT COLONOGRAPHY USING MATHEMATICAL MORPHOLOGY Mozhdeh Shahbazi a, *, Mehran Sattari b , Mojtaba Ghazi c a Dept. of Surveying, Engineering Faculty, University of Isfahan, Isfahan, Iran- n_shahbaz2003@yahoo.com b Dept. of Surveying, Engineering Faculty, University of Isfahan, Isfahan and Dept. of Surveying, Engineering Faculty, University of Tehran, Tehran, Iran- Sattari@eng.ui.ac.ir c Alzahra Educational Hospital , Medical Science University of Isfahan, Isfahan, Iran- mojtaba_ghazi_t@yahoo.com Working Group Sessions and related Poster Sessions, WgS – PS: WG V/6 KEY WORDS: Image Processing, Detection, Computer vision, Feature Detection, Medicine, Feature Extraction, Feature Recognition ABSTRACT: In this paper we present and develop a set of algorithms, mostly based on morphological operators, for automatic colonic polyp detection applied to computed tomography (CT) scans. Initially noisy images are enhanced using Morphological Image Cleaning (MIC) algorithm. Then the colon wall is segmented using region growing followed by a morphological grassfire operation. In order to detect polyp candidates we present a new Automatic Morphological Polyp Detection (AMPD) algorithm. Candidate features are classified as polyps and non-polyps performing a novel Template Matching Algorithm (TMA) which is based on Euclidean distance searching. The whole technique achieved 100% sensitivity for detection of polyps larger than 10 mm and 81.82% sensitivity for polyps between 5 to 10 mm and expressed relatively low sensitivity (66.67%) for polyps smaller than 5 mm. The experimental data indicates that our polyp detection technique shows 71.73% sensitivity which has about 10 percent improvement after adding the noise reduction algorithm. * Corresponding author. 1. INTRODUCTION Colon cancer death is among increasing causes of death (Jemal et al., 2004). Most colorectal cancer mortalities can be prevented by early detection and removal of colonic polyps (Robert Van Uiterta et al., 2006). A way to diagnose colonic polyps is to screen colon via colonoscopy. Figure 1 is a digital photograph from conventional colonoscopy showing a colonic polyp. Although colonoscopy provides a precise means of colon examination, it is time-consuming, expensive to perform, and requires great care and skill by the examiner. Moreover, since colonoscopy is an invasive procedure, there is a fatal risk of injury to colon. In comparison with colonoscopy, Computed Tomography scanning is a technique for non-invasively performing colon cancer screenings. According to radiologists, it is not that simple to distinguish colon wall and successively colonic polyps from CT slices. Therefore, automatic polyp detection can make diagnostic processes reach a general level, not depending highly on the experts' special skills. In this regard, Vining et al., 1997 proposed a method to detect the colonic polyps by analysing the local curvature of the colon surface attaining 73% sensitivity. Summers et al., 2001 developed a method that identifies the convex surfaces that protrude inward from the colon by evaluating the principle and mean curvature of the colon surface. Their method achieved 29% to 100% sensitivity. Yoshida et al., 2002 proposed to use features such as the shape index (cup, rut, saddle, ridge, and cap) and curvedness values on small volume of interest and apply fuzzy clustering for polyp detection. They reported 89% sensitivity. Paik et al., 2000 proposed a technique based on contour normal intersection to detect surface patches along the colon wall and shows 85% to 90% sensitivity. Kiss et al., 2002 combined the surface normal distribution and sphere fitting to produce 90% polyp sensitivity for polyps higher than 6mm. Kiss et al., 2003 employed the slope density function to discriminate between polyps and folds and their technique shows 85% sensitivity for polyps higher than 6mm. Paik et al., 2004 developed a new technique based on surface normal overlap. Acar et al., 2001 suggested a method that detects spherical patches by Hough Transform and the algorithm analyses them using the optical flow to decide if they are polyps or not. Other interesting automated CAD techniques include the work of Gokturk et al., 2001, Acar et al., 2002, Wang et al., Figure 1. Colonic polyp from conventional colonoscopy 823