Artificial intelligence in surveillance of barrett's esophagus

Abstract

A study by Waterhouse and colleagues in a previous issue of Cancer Research describes the development and prospective validation of an artificial intelligence approach in conjunction with spectral imaging to enhance endoscopic detection of Barrett's esophagus-related neoplasia. The authors developed a novel spectral endoscope with external optics suitable for routine Barrett's esophagus surveillance with diffuse tissue reflectance to define multispectral data correlated with histopathology. A convolutional neural network was trained on the absis of the spectral signatures acquired as part of a small, prospective clinical trial to distinguish Barrett's esophagus from Barrett's esophagus neoplasia. The results from the study suggest the utility of artificial intelligence for diagnosis of Barrett's esophagus.