Open-Source Datasets for Colonoscopy Polyps and Its AI-Enabled Techniques

High-quality, open access, and free colonoscopy data act as a catalyst for the ongoing state-of-the-art (SOTA) artificial intelligence (AI) research works in the detection, localization, segmentation, and classification of various colorectal cancer findings in the colon and rectum such as polyp growth, abnormal tissues, cancerous and non-cancerous lesions, etc. This paper presents and summarizes widely used, open-source, and downloadable polyp colonoscopy datasets along with its source links. A comparative analysis of twenty-one colonoscopy datasets has been done. Copy-right-free, raw colonoscopy data (videos and images) from various medical Websites and YouTube videos are also mentioned for potential data development. Such colonoscopy datasets will further aid in the development of AI-powered software and hardware medical systems in this field and eventually help in reducing the burden of experienced gastroenterologists.