Optical cholangioscopic characterization of biliary strictures is essential to predict if a given stricture is of malignant or non-malignant etiology. Nevertheless, this characterization is has significant interobserver variability.
Our group developed an artificial intelligence algorithm based on convolutional neural networks for automatic detection of papillary projections in digital cholangioscopy images (Spyglass DS II). Our system detected this cholangioscopic feature with a sensitivity of 100% and a specificity of 97% Its overall accuracy was 98%. Application of artificial intelligence to digital cholangioscopy may improve the accuracy of visual impression diagnosis as well as improve its significant interobserver variability. The link for the complete article will posted here soon.

Accepted Manuscript // August 2021