A new approach to detecting ulcer and bleeding in Wireless capsule endoscopy images.

In recent years, Wireless capsule endoscopy (WCE) has been widely utilized in diagnosis of gastrointestinal (GI) tract disease. This new technology is painless and can see small intestine that traditional endoscopies cannot reach. However, Analysis of massive images for each WCE detection is tedious and time consuming to physicians. In this paper we present a computer-aid approach to help clinicians to discriminate amongst regions of normal or abnormal tissue. We use covariance of second-order statistical features which called as color wavelet covariance (CWC), based on discrete wavelet transform (DWT) and then optimize them by a selected algorithm. Accurate image segmentation and classification is achieved by a joint classifier, which is obtained by Texton Boost classifier. The whole approach has been validated on various WCE data and achieves a good result.