Dalia Braverman-Jaiven Luigi Manfredi ^*

• School of Medicine, University of Dundee, Dundee, United Kingdom

Inflammatory bowel disease (IBD) causes chronic inflammation of the colon and digestive tract, classified as Crohn's disease (CD) and Ulcerative colitis (UC). IBD is more prevalent in Europe and North America, but since the 21st century, it has been increasing in South America, Asia, and Africa, making it a global concern. Optical colonoscopy is crucial for diagnosing and assessing IBD, providing real-time visualization of the colonic wall and ileum and enabling tissue sample collection. The accuracy of colonoscopy depends on the endoscopist's expertise, thus algorithms based on Deep Learning (DL) and Convolutional Neural Networks (CNN) for colonoscopy images and videos are increasingly popular, especially for detecting and classifying colorectal polyps. The system's performance relies on the quality and quantity of training data. While several datasets are available for endoscopy images and videos, most specialize in polyps. The use of DL algorithms to detect IBD is still in its early stages, with most studies focused on assessing UC severity. As AI gains popularity, interest in using these algorithms for diagnosing and classifying IBDs and managing their progression is growing. To address this, more annotated colonoscopy images and videos are needed for training new, reliable AI algorithms. This article discusses the current challenges in early IBD detection, focusing on available AI algorithms and databases and the challenges ahead to improve detection rates.