- Published literatures have reported that 26% of diminutive neoplastic colorectal polyps are missed in a single colonoscopy. This may be due to blind spots or human error during colonoscopy. Artificial intelligence based colorectal polyp detection may improve the polyp detection. In this study, we aim to develop newer models of artificial intelligence (AI) methods to improve accuracy in detection of malignant diminutive colorectal polyps.
- The gold standard and most widely used approach for screening and surveillance of Barrett’s esophagus (BE) is esophagogastroduodenoscopy. However, the visual detection of early esophageal neoplasia (high grade dysplasia and T1 stage adenocarcinoma) in BE with white light and electronic virtual chromoendoscopy is still often difficult. The aim of this study is to assess if a convolutional neural artificial intelligence network can aid in the recognition of early esophageal neoplasia in BE.
- Adenoma detection rate (ADR) is a primary quality indicator for colonoscopy based on its inverse relationship with interval colorectal cancers. The quality of colonoscopy preparation, as determined by validated scoring systems, is directly related to ADR and is a determinate for timing of next colonoscopy. Among scoring systems, Boston Bowel Preparation Score (BBPS) has gained favor as it reflects an “achieved prep” after appropriate washing and aspiration. Yet, agreement between endoscopists is imperfect (kappa 0.74) and BBPS does not account for small segments within scored regions.
- Sessile serrated adenomas/polyps (SSA/Ps) should be resected because they have potential to be cancer, however endoscopic differentiation of SSA/Ps from hyperplastic polyps or flat adenomas with endoscopists’ eyes is considered difficult. In this pilot study, we evaluated the performance of the newly developed artificial intelligence (AI) in endoscopic identification of SSA/Ps.