Artificial intelligence in endoscopy-new ways to detect and characterize polyps

Screening colonoscopy is the most effective strategy for prevention of colorectal cancer, but the efficacy is highly dependent on the quality of the endoscopic procedure. The most important determinant factor is the detection, characterization and removal of all polyps and premalignant adenomas. The rate of so-called interval carcinoma correlates directly with the percentage of missed adenomas during the screening procedure. Automated polyp detection algorithms can improve and optimize the detection and characterization of colorectal polyps. The possibilities of image analysis methods and structure analysis have improved enormously in recent years mainly by the use of artificial intelligence (AI) implementing convoluted neuronal network and big data analysis. The current developmental activities are directed, on the one hand, toward automated detection of polyps during live endoscopy and, on the other hand, to identify and characterize the type and histology of an identified polyp in the context of a computer-aided diagnosis (CAD). The current review provides a summary of the different approaches and software solutions which are currently available to improve screening colonoscopy.