Development and implementation of artificial intelligence for better detection and characterization of colorectal diseases during colonoscopy

Colorectal cancer is one of the most frequent cancers in the Western world. Many countries have implemented screening programs that include colonoscopy. The success depends on the quality of the procedure which is very operator dependent. Indeed, patients that undergo a colonoscopy by an endoscopist with a low polyp detection rate are at greater risk to still develop colorectal cancer despite undergoing colonoscopy. We aim to develop and validate an artificial intelligence (AI) tool to improve polyp detection. Screening programs are also associated with an increased cost, associated with the removal and microscopic analysis of polyps that are often found. The majority of these small polyps are however completely benign. We aim to develop an AI tool for optical diagnosis that holds the potential to make a diagnosis based on the endoscopic image and can be cost saving for the society if performed with high accuracy. Likewise, disease assessment in ulcerative colitis is operator dependent, but it would be useful to find out which patients can stop expensive and possibly harmful treatment. We want to further clinically validate two AI tools (Red Density and single wavelength endoscopy) to assess better mucosal and histological healing and predict long term remission. AI has entered daily life. Also in medicine, it has the potential to aid physicians in patient management, improve quality of care and be cost saving. In a multidisciplinary collaboration with imaging engineers, we will further develop and clinically implement our AI systems to improve polyp detection and characterisation and to quantify inflammation.