Loading...
Developments are accelerating in the application of artificial intelligence (AI) systems to aid polyp detection and characterization during colonoscopy. Convolutional neural networks (CNNs) classify images using algorithms to identify many aspects of visual data. In this study, researchers used more than 8600 screening colonoscopy photographs to train a deep CNN to detect polyps. The CNN's efficacy was tested on a data set of 1330 images, and expert colonoscopists (adenoma detection rate, ≥50%) were asked to identify, with or without CNN overlay, polyps in 9 colonoscopy videos of 28 polypectomies. The results were as follows:
CNN accuracy for polyp detection in frames was greater than 96%.
Processing time was 10 milliseconds per frame
Expert r…