Implementation of Artificial Intelligence in Colonoscopic Diagnostics

Last decade artificial intelligence (AI) has become our whip hand both in daily life and in medicine. This project focusses on the implementation of AI during colonoscopy.

Screening for colorectal cancer, the second most frequent cancer in the general population, is crucial in the prevention of cancer related mortality. AI may better detect and distinguish benign from premalignant lesions and therefore helps to decide which polyp to resect and to make a histological diagnosis without the costs of classical pathology.

An in-house trained AI tool indicates detected polyps during real-time endoscopy and will be validated in a multicenter prospective trial. Features as polyp characterization and delineation in relation with the histopathology will be studied in close cooperation with engineers of the MIRC and clinicians aiming for a quality improvement of future endoscopy.

Ulcerative colitis (UC) is a bowel disease characterized by a chronic inflammation of the colon. When uncontrolled there is a risk of losing the colon due to persistent inflammation. The therapy’s goal is clinical and mucosal healing, the latter requiring endoscopic evaluation. I will prospectively validate new endoscopic techniques able to quantify endoscopic healing with a good correlation to histology. The feasibility and prognostic potential will be investigated as it provides an independent objective technique for endoscopic evaluation of inflammation in UC and prediction of disease relapse.