Automatic Bleeding Frame and Region Detection for GLCM Using Artificial Neural Network

Abstract

 Wireless capsule endoscopy is a device that inspects the direct visualization of patient’s gastrointestinal tract without invasiveness. Analyzing the WCE video is a time- consuming task hence computer aided technique is used to reduce the burden of medical clinicians. This paper proposes a novel color feature extraction method to detect the bleeding frame. First, we perform word based histogram for rapid bleeding detection in WCE images. Classification of bleeding WCE frame is performed by applying for glcm using  Artificial Neural Network and K-nearest neighbour method. Second we propose a two-stage saliency map extraction method. In first stage saliency, we inspect the bleeding images under different color components to highlight the bleeding regions. From second stage saliency red color in the bleeding frame reveals that the region is affected. Then, by using algorithm we fuse the two-stage of saliency to detect the bleeding area. Experimental results show that the proposed method is very efficient in detecting the bleeding frames and the region.