Colorectal cancer is the fourth most common cause of cancer death worldwide, with about 143.460 new cases expected in 2012 by recent estimates of the American Cancer Society. Colon cancerís survival rate depends on the stage in which it is detected, decreasing from rates higher than 95% in the first stages to rates lower than 35% in stages IV and V, hence the necessity for a early colon screening. In this process physicians search for adenomatous growths known as polyps, in order to assess their degree of development. There are several screening techniques but colonoscopy is still nowadays the gold standard, although it has some drawbacks such as the miss rate. Our contribution, in the field of intelligent system for colonoscopy, aims at providing a polyp localization and a polyp segmentation system based on a model of appearance for polyps. In this sense we define polyp localization as a method which given an input image identifies which areas of the image are more likely to contain a polyp. Polyp segmentation aims at selecting the region of the image that contains a polyp. In order to develop both methods we have started by defining a model of appearance for polyps, which defines a polyp as enclosed by intensity valleys. The novelty of our contribution resides on the fact that we include in our model other elements from the endoluminal scene such as specular highlights and blood vessels, which have an impact on the performance of our methods and also other elements that appear as a result of image formation, such as interlacing. Considering this we define our novel Depth of Valleys image which integrates valley information with the output of the morphological gradient and also takes into account the presence of the before mentioned elements of the endoluminal scene. In order to develop our polyp localization method we accumulate the information that the Depth of Valleys image provides in order to generate accumulation energy maps. In order to obtain polyp segmentation we also use information from the energy maps to guide the process. Our methods achieve promising results in polyp localization and segmentation. In order to validate our methods we also present an experiment which compares the output of our method with physicianís observations captured via an eye-tracking device. The results show to be close to physicianís observations which point out a potentially inclusion of our methods as part of a future intelligent system for colonoscopy.
Polyp Localization and Segmentation in Colonoscopy Images by Means of a Model of Appearance for Polyps
Bernal Del Nozal, J. (Author). 17 Dec 2013
Student thesis: Doctoral thesis
Student thesis: Doctoral thesis