Endoscopy is considered the gold standard for the detection of gastrointestinal (GI) disease and malignancies. However, the diagnostic accuracy highly depends on the quality of the endoscopic procedure. The most important quality indicator for colonoscopies, the Adenoma Detection Rate (ADR), has proven to be highly dependent on the ability of the endoscopist to adequately visualize the mucosa. Furthermore, it is known that up to 11% of upper GI malignancies have been missed during gastroscopy. Improvement of the diagnostic yield of endoscopy is urgently needed as this will lead to fewer missed diagnoses and thereby earlier treatment, improvement of patient outcomes and treatment costs.
It is suggested that artificial intelligence (AI) can improve the quality of endoscopies. Several studies regarding Computer-Aided Detection (CADe) already have been published for the detection of specific lesions such as polyps. However, complete and adequate mucosal visualization is a prerequisite for the detection of pathology, especially for smaller and more subtle lesions. Therefore, we aim to develop a real-time AI-based tool to optimize mucosal visualization during endoscopy.
The project is in collaboration with the Department of Gastroenterology and Hepatology, Radboudumc.