UEG's content platform for gastroenterology education bringing recordings, courses, guidelines, and more together in one place.

UEG Week Berlin 2025 Recordings Gut Guide online What's new on Gutflix

Trending on Gutflix

Rate for better recommendations!

Your feedback helps us show you more of what you like and less of what you dislike!

More of this!

We will show you more similar content.

Not for me!

We won't suggest this again.

Bookmark!

Save this content for later!

New to myUEG? Create an account! Log In with myUEG account

UEG Poster

Audio / Video Poster

Share via Email Share on Facebook Share on X Share on LinkedIn Share on Bluesky

This content is part of Gutflix. Log in with your myUEG account, or create one free, to watch it.

Not sure what you can access? Learn more about account types.

APPLICATION OF ARTIFICIAL INTELLIGENCE IN THE DETECTION OF BORRMANN TYPE 4 ADVANCED GASTRIC CANCER IN UPPER ENDOSCOPY

Mi Jin Oh 1, Jinbae Park 2, Jiwoon Jeon 2, Sang-kyun Ko 2, Seyeong Jo 2, Seungkyung Kang 3, Su-Hyun Kim 3, Su Hee Park 3, Young Hoon Chang 4, Cheol-Min Shin 4, Seung Joo Kang 5, Sang Gyun Kim 1, Soo-Jeong Cho 1

Affiliations

1 Seoul National University Liver Research Institute, Seoul, Korea, Republic of

2 Ainex Co., LTD., Seoul, Korea, Republic of

3 Seoul National University Hospital, Seoul, Korea, Republic of

4 Seoul National University Bundang Hospital, Seongnam-si, Korea, Republic of

5 Healthcare System Gangnam Center, Seoul National University Hospital, Seoul, Korea, Republic of

Summary

AI Generated

An artificial intelligence system using ResNet-152 was developed to detect Borrmann type 4 gastric cancer from upper endoscopic images, achieving 92.12% accuracy on image-level identification and 97% accuracy on patient-level diagnosis.

Borrmann type 4 gastric cancer is a diffuse gastric cancer that infiltrates the gastric wall without overt mass or ulceration, is associated with poor prognosis, and is easily missed on routine endoscopy.
The AI model was trained on 227 gastric cancer patients and 227 controls from Seoul National University Hospital using a ResNet-152 deep learning network.
Image-level performance showed 87.29% sensitivity, 97.97% specificity, and 92.12% accuracy for identifying Borrmann type 4 gastric cancer.
Patient-level diagnosis using a threshold of more than 50% positive images improved performance to 94% sensitivity, 100% specificity, and 97% accuracy.
The authors note that further internal and external validation is needed before clinical implementation.

Download PDF

Was this helpful?

Thanks for your feedback.

This summary was generated by an AI large language model based on the content transcript. It is for informational purposes only and should not be considered a substitute for clinical judgment. Always rely on your professional expertise and the full clinical context when making clinical decisions.

Abstract

Introduction

Borrmann type 4 gastric cancer is defined as a diffuse gastric cancer that infiltrates the gastric wall, often without an overt mass or ulceration. Clinically, it is associated with a poor prognosis and lower curative resection rate, but is easily missed in routine upper endoscopic exams. This study aimed to develop an artificial intelligence (AI) system that can detect Borrmann type 4 gastric cancer lesions in upper endoscopic images.

Aims & Methods

Endoscopic images of patients diagnosed with Borrmann type 4 gastric cancer, along with normal controls were collected from Seoul National University Hospital. A total of 227 gastric cancer patients (1,973 images) and 227 normal controls (1,554 images) were assigned to the train dataset, while 50 gastric cancer patients (417 images) and 50 normal controls (344 images) were assigned to the test dataset. A deep residual learning network model, ResNet-152 was used for training.

Results

The sensitivity, specificity and accuracy for identification of Borrmann type 4 gastric cancer images were 87.29% (95% confidence interval [CI] 84.09-90.49%), 97.97% (95% CI 96.47-99.46%) and 92.12% (95% CI 90.20-94.03%), respectively. When the diagnosis was made in patients with more than 50% of the endoscopic images identified as Borrmann type 4 gastric cancer, the sensitivity, specificity and accuracy was improved to 94% (95% CI 87.42-100%), 100% (95% CI 100-100%) and 97% (95% CI 96.66-100%), respectively.

Conclusion

A novel AI-based model that can identify Borrmann type 4 gastric cancer from endoscopic images was developed. Although further internal and external validation is needed, it is expected to assist clinicians in the early diagnosis of Borrmann type 4 gastric cancer with high accuracy.

Conference

UEG Week Copenhagen 2023

Topics

Oesophagus

Submission format

Abstract

Session

PP 01 Oesophageal, gastric and duodenal disorders (Posters)

Citation

United European Gastroenterology Journal 2023; 11 (Supplement 8)

Published

2023

More Like This:

The guide to infection-free endoscopy in UZ Ghent

Thalita Vanhullen

Care of psychiatric patients during endoscopic treatment of repeated foreign body ingestions

Karolien Haenebalcke

When brushing teeth becomes a challenge: Foreign body removal

Arianna Parrella

Education of endoscopy nurses in Europe – 30 years and still not standardised

Ulrike Beilenhoff

CLINICAL PHENOTYPES THROUGH MACHINE LEARNING OF FIRST-LINE TREATED PATIENTS DURING THE PERIOD 2013-2022: DATA FROM THE EUROPEAN REGISTRY ON HELICOBACTER PYLORI MANAGEMENT (HP-EUREG)

Samuel Jesús Martínez-Domínguez

CLINICOPATHOLOGICAL FEATURES AND SERUM RISK MARKERS OF METACHRONOUS GASTRIC CANCER AFTER ERADICATION

Ken Yamashita

UEG Presentation

New

Share via Email Share on Facebook Share on X Share on LinkedIn Share on Bluesky

The guide to infection-free endoscopy in UZ Ghent

This content is part of Gutflix. Log in with your myUEG account, or create one free, to watch it.

Not sure what you can access? Learn more about account types.