Artificial Intelligence-Based Colorectal Polyp Histology Prediction by Using Narrow-Band Image-Magnifying Colonoscopy

• DOI: 10.5946/ce.2021.149
• Publication Date: 2021-09-23T07:25:53Z
• Authors: Istvan Racz; Andras Horvath; Noemi Kranitz; Gyongyi Kiss; Henriett Regoczi; Zoltan Horvath
• Keywords: Histology; Hyperplastic Polyp; Diminutive; Narrow-band imaging; Colorectal Polyp; Polypectomy

We have been developing artificial intelligence based polyp histology prediction (AIPHP) method to classify Narrow Band Imaging (NBI) magnifying colonoscopy images predict the hyperplastic or neoplastic of polyps. Our aim was analyze accuracy AIPHP and narrow-band imaging international colorectal endoscopic (NICE) classification predictions also compare results two methods.We studied 373 samples taken by polypectomy from 279 patients. The documented NBI still were analyzed NICE parallel. software created machine learning method. measures five geometrical color features on image.The 86.6% (323/373) in total compared for diminutive non-diminutive polyps (82.1% vs. 92.2%; p=0.0032). significantly better both (n=207) (95.2% 82.1%) (p<0.001) all evaluated (n=373) (97.1% 86.6%) (p<0.001).Our could with high only large size subgroup.