The Newest Tech for Your Practice | August 2021-GIE Journal Article Neural Network

Pdf Summary

Researchers have developed a convolutional neural network (CNN) that can predict the histology of early squamous cell cancer of the esophagus in real time based on intrapapillary capillary loop (IPCL) patterns. The CNN was trained using magnification endoscopy narrow-band imaging videos of squamous mucosa and achieved a diagnostic performance comparable to an expert panel of endoscopists, with an average F1 score of 94%, sensitivity of 93.7%, and accuracy of 91.7%. The CNN classified IPCL patterns at video rate and generated class activation maps that visually validated its predictions. The use of CNNs in diagnostic endoscopy is gaining momentum, and this study contributes to the growing body of research on the application of artificial intelligence in the diagnosis of early cancers. Further development of the CNN could include predicting the depth of histologic invasion and improving the detection of abnormal esophageal mucosa. The CNN has the potential to be a useful diagnostic adjunct for endoscopists involved in the management of early squamous cell neoplasia of the esophagus, particularly in settings where experience or training may be limited.

Asset Subtitle

A clinically interpretable convolutional neural network for the real-time prediction of early squamous cell cancer of the esophagus: comparing diagnostic performance with a panel of expert European and Asian endoscopists.

Meta Tag

Disease Early Squamous Cell Neoplasm (ESCN)

Instrument & Accessory Used Narrow Band Imaging (NBI)

Instrument & Accessory Used Magnification Endoscopy

Organ & Anatomy Esophagus

Organ & Anatomy Stomach

Procedure EGD

Keywords

convolutional neural network

histology

early squamous cell cancer

esophagus

intrapapillary capillary loop

IPCL patterns

magnification endoscopy narrow-band imaging

diagnostic performance

artificial intelligence

diagnosis of early cancers