Computer aided detection of bleeding in capsule endoscopy images

The capsule endoscopy (CE) has been widely used to diagnose the diseases in human digestive tract because of its great breakthrough that it can view the entire small bowel without invasiveness. However, a tough problem associated with this new technology is that too many images to be examined by eyes cause a huge burden to physicians, so it is very useful to help the physician do diagnosis using computerized methods. In this paper, a new method aimed for bleeding region detection in CE images is proposed. This new approach mainly focuses on color texture feature, also a very important clue for the physicians to judge the status of the gastrointestinal tract. We propose a new idea of chromaticity moment as the color feature, which make full use of the Tchebichef polynomials and the illumination invariant of the HSI color space. Then, combined with the uniform local binary pattern (LBP), a traditional texture representation model, it can be used to discriminate normal regions and bleeding regions. Classification of bleeding regions with multilayer perceptron neural network is then deployed to verify the performance of the proposed new color texture features. Experimental results on our present bleeding image data sets show that this new scheme is promising in detecting bleeding regions.