Principal Curvature Based Polyp Detection in Wireless Capsule Endoscopy Images

Polyp identification in human Gastro Intestinal (GI) Tract employing endoscopy and colonoscopy through various computer vision and image processing techniques has achieved lot of success and maturity in its field. Over the last decade Wireless Capsule Endoscopy (WCE) has emerged as an innovative diagnostic system for identification of ailments such as polyps in GI tract. Polyp detection through WCE suffers from various types of challenges due to its unique image acquisition technique. In the proposed work, Principal curvature based polyp detection technique in WCE is presented. The experimental results were conducted on 58 WCE images comprising of various types of polyps. The results have revealed that an adaptive geometrical feature extraction based identification method of polyps is suggested for efficient analysis. Further, the challenges associated with detection of polyps are also discussed.