New AI model accurately classifies colorectal polyps using slides from 24 institutions

(Dartmouth-Hitchcock Medical Center) Researchers at Dartmouth's and Dartmouth-Hitchcock's Norris Cotton Cancer Center have trained a deep neural network to distinguish the four major types of colorectal polyps excised during screening colonoscopy. The model not only produced results that demonstrated accuracy and sensitivity at the level of practicing pathologists, but withstood evaluation using broad datasets spanning multiple institutions across the US. The study proves that AI models are generalizable and can be trained on widespread external data.
Source: EurekAlert! - Cancer - Category: Cancer & Oncology Source Type: news