Bleeding fragment localization using time domain information for WCE videos

Small intestine bleeding is one of the most common disease in human gastrointestinal tract. The Wireless Capsule Endoscopy (WCE) has been an advanced technology to inspect the intestine in recent years. In this paper, a new time domain based method was proposed to locate bleeding fragment in WCE videos. The spatial feature and the time domain information were combined to make the decision. The proposed method consists of four phases: First, dark holes and wrinkle edges of each frame were removed in preprocessing. Second, each frame was segmented into small patches by using super-pixel segmentation. A coarse bleeding region filter was adopted to remove noisy patches. Third, the red ratio feature and the a feature of the WCE images were extracted. Finally, the bleeding fragment was detected based on the temporal red ratio feature and the temporal a features fusion. Experimental results show that the proposed method achieves good performance.