Optical diagnosis of colorectal polyps, the in-vivo characterization of the histology by endoscopists, is of increasing interest for clinical endoscopy practice. Recent studies have shown that thresholds for optical diagnosis are met in highly selected groups of expert endoscopists, but the same is not true in community endoscopy practices. In order to improve optical diagnosis, imaging enhancement techniques and the use of artificial intelligence are proposed.
This observational study developes a computer-aided diagnosis (CADx) system to differentiate between benign and (pre-)malignant CRPs, using state-of-the-art machine learning methods and deep learning architectures. For the development, HDWL and BLI images are used. The CADx is trained using histology as gold standard. The CADx is externally validated using on a set of 60 colorectal polyps. This study will evaluate if the optical diagnosis of colorectal polyps can be improved with the aid of CADx.
In addition, both expert endoscopists and novices optically diagnose the colorectal polyps. In the first, pre-training phase, endoscopists optically diagnose colorectal polyps based on intuition. Afterwards, in the post-training phase, the same set of colorectal polyps is optically diagnosed based on a clinical classification system; BLI Adenoma Serrated International Classification (BASIC). This study will evaluate if the optical diagnosis of colorectal polyps can be improved with the aid of BASIC in both expert and non-expert hands.
