Bleeding region detection in WCE images based on color features and neural network

Wireless Capsule Endoscopy (WCE) is the first-line procedure in Obscure Gastrointestinal Bleeding (OGIB). However, a major problem associated with this new technology is that too many images need to be manually examined by clinicians and the integrated software SBI has insufficient sensitivity and specificity. In this paper a method aimed for rapid detecting bleeding regions in the WCE video is proposed. As color features are most important clues for the clinician to judge the status of the gastrointestinal tract, the proposed technique mainly focuses on color features in RGB color spaces. The results obtained from comparative experiments on our image data sets show that this new technique is superior in detecting bleeding frames in the WCE video.