Endoscopy: Computer-Aided Diagnostic System Based on Deep Learning Which Supports Endoscopists’ Decision-Making on the Treatment of Colorectal Polyps

The quality and quantity of endoscopic images that physicians obtain are dramatically increasing along with the recent advance in imaging technologies. However, benefits coming from these rich data may not be effectively utilized by all the endoscopists due to the lack of shared knowledge and experience. The use of artificial intelligence (AI) as a decision support during endoscopy is catching great attention as a measure to overcome this issue. In the colonoscopy field, AI is expected to facilitate polyp detection, prediction of polyp pathology, and prediction of invasion depth of colorectal cancer. With the use of AI technologies, the macroscopic anatomy (i.e., endoscopic view) is matched or fused with microscopic findings (i.e., pathological finding) in real time during the endoscopic examination. This new methodology allows clinical doctors to make a decision much easier than the conventional method. These research concepts and results well fit in the anatomy-pathology axis of the multidisciplinary computational anatomy (MCA) model.