An Automatic Colonic Polyp Type Identification System by Narrow-Band Imaging and Focal Zone Features of Vascular Shapes and Patterns

Given changing dietary habits and increased consumption of high-calorie foods, colorectal cancer has become a widespread disease. When colorectal cancer symptoms are ignored, a carcinoma can develop. Therefore, physicians must diagnose colorectal cancer in a timely manner. Unlike endoscopic colonoscopy, narrow-band imaging improves identification of non-neoplastic and neoplastic colonic polyps by recognizing the shapes and patterns of blood vessels. This work identifies different colonic polyp types via narrow-band imaging based on focal zone features of vascular shapes and patterns. The proposed system can identify different polyp types via selection of focal zone images, feature extraction of vessels, and a decision model.