Related Experiment Video
Updated: Feb 16, 2026
05:33
Introduction of an Integrated Pathology Image Management, Artificial Intelligence, and Reporting System
Published on: July 11, 2025
1.2K
Computer-Aided Diagnosis Based on Convolutional Neural Network System for Colorectal Polyp Classification:
Yoriaki Komeda1, Hisashi Handa, Tomohiro Watanabe
1Department of Gastroenterology and Hepatology, Kindai University Faculty of Medicine, Osaka-Sayama, Japan.
Oncology
|December 20, 2017
Summary
A novel artificial intelligence (AI) system using convolutional neural networks (CNN) demonstrated preliminary success in diagnosing colon polyps from routine colonoscopy images. This AI-powered computer-aided diagnosis (CAD) tool shows potential for improving polyp classification accuracy.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Gastroenterology
Background:
- Computer-aided diagnosis (CAD) is evolving for disease detection.
- CAD for colon polyps aids trainee colonoscopists by reducing risks associated with endoscopic resections.
- Convolutional Neural Network (CNN) systems utilizing artificial intelligence (AI) have advanced rapidly.
Purpose of the Study:
- To develop a unique CNN-CAD system with AI capabilities for diagnosing colon polyps.
- To train the AI system using endoscopic images from routine colonoscopies.
- To report preliminary findings of this novel CNN-CAD system.
Main Methods:
- Utilized 1,200 images from colonoscopies performed between January 2010 and December 2016.
- Extracted images from actual endoscopic examination videos.
- Trained the AI on 1,200 images (600 adenomatous, 600 nonadenomatous) adjusted to 256x256 pixels.
- Conducted a 10-fold cross-validation.
Main Results:
- Achieved an accuracy of 0.751 in the 10-fold cross-validation.
- The CNN correctly classified polyps in 7 out of 10 cases during pilot assessment.
- The AI system learned to distinguish between adenomatous and nonadenomatous polyps.
Conclusions:
- A CNN-CAD system applied to routine colonoscopy images shows promise for rapid colorectal polyp classification.
- Further in vivo prospective studies are necessary to validate the effectiveness of this CNN-CAD system in clinical practice.
