Loading...
There has been what can only be called an explosion of interest in applying artificial intelligence (AI) to gastrointestinal endoscopy. Using machine learning algorithms, computers can be taught to identify and classify colonic polyps, Barrett esophagus, gastric cancer, and other gastrointestinal lesions. In a recent study, researchers used AI to distinguish autoimmune pancreatitis (AIP) from pancreatic cancer, a distinction that can sometimes vex even an experienced endosonographer.
The investigators developed a convolutional neural network (CNN) using endoscopic ultrasound (EUS) images and videos of normal pancreas (NP), chronic pancreatitis (CP), and AIP. Over 1 million unique images from 583 patients were evaluated. In an analysis using …