AN INTELLIGENT WEB ATLAS IN DIGESTIVE ENDOSCOPY

摘要

Faced with an ambiguous situation, the diagnostic decision of the endoscopist could be improved by the fast access to similar images extracted from a referring indexed database. The purpose of this paper is to present an intelligent atlas, based on a reasoning model of the diagnostic decision in digestive endoscopy. The observation of endoscopists and endoscopic terminologies suggests that the diagnostic reasoning process in digestive endoscopy uses a scene - object approach. The objects correspond to the endoscopic findings. The scene corresponds to the endoscopic diagnosis and includes the medical context of the examination. According to this reasoning model, classes of endoscopic findings and diagnoses are described in an intensive way by the symbolic strength of their relationship with indices. The retrieval method is based on a typicality metric that estimates the closeness between the case under investigation and the prototype of the class. At the present time, the system contains 150 endoscopic images and is accessible via Internet. The next step will integrate the system in a learning tool for junior endoscopists.