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

Re-live UEG Week 2026 on Gutflix 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

Standard 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.

AUTOMATICALLY OPTIMIZED RADIOMICS MODELING SYSTEM FOR SMALL GASTRIC SUBMUCOSAL TUMORS (&LT;2CM) DISCRIMINATION BASED ON ENDOSCOPIC ULTRASOUND IMAGES

Mingyan Cai 1, Baohui Song 1, Pingting Gao 1, Yunshi Zhong 1, Yinhui Deng 2, Jinhua Yu 2, Pinghong Zhou 1

Affiliations

1 Zhongshan Hospital, Fudan Univeristy, Shanghai, China

2 Fudan University, Shanghai, China

Summary

AI Generated

A machine learning model using endoscopic ultrasound images demonstrated stable performance for diagnosing gastrointestinal stromal tumors and stratifying malignant risk in small gastric submucosal tumors under 2cm, outperforming experienced endoscopists in a prospective multicenter validation.

The automatically optimized radiomics modeling system achieved an AUC of 0.770 and accuracy of 0.742 for diagnosing GIST versus non-GIST in a prospective validation cohort of 178 patients from different centers.
For risk stratification distinguishing group 1 and group 4 GIST, the model achieved an AUC of 0.750 and accuracy of 0.872 in the validation set.
The model outperformed five experienced endoscopists whose diagnostic accuracy ranged from 0.449 to 0.531 for GIST diagnosis and 0.702 to 0.809 for risk stratification.
The development cohort included EUS images from 205 small gastric submucosal tumor patients under 2cm analyzed using three-fold cross-validation.
This work is most relevant to endoscopists and gastroenterologists managing small gastric submucosal tumors requiring differentiation of GIST from other entities and assessment of malignant potential.

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

Distinguishing gastrointestinal stromal tumors (GISTs) with high malignant risk is essential for the management of small submucosal tumors (SMTs).

Aims & Methods

We aimed to develop a machine learning model based on endoscopic ultrasound (EUS) images to diagnose GISTs and non-GISTs as well as distinguish their potential malignancy. EUS images of 205 small gastric SMT (<2cm) patients were retrospectively included in the model development phase. An automatically optimized radiomics modeling system (AORMS) framework was developed to carry out the model learning based on the features extracted from the retrospective cohort. Three-fold cross-validation was used to train the machine learning model and produce the optimal model for differentiating between GIST and non-GIST as well as groups 1 and 4 GIST. Images of 178 patients from different centers were prospectively enrolled in the validation phase of the model. The diagnostic performance of the AORMS was evaluated in the development set and compared with the performance of endoscopists. Then, the performance of the AORMS was evaluated in the prospective validation set.

Results

The overall area under the curve (AUC), accuracy, sensitivity, and specificity of the AORMS were 0.762, 0.708, 0.819 and 0.649 for the diagnose of GIST, and 0.734, 0.872, 0.571 and 0.908 for the risk stratification of group 1 and 4 GIST. The accuracy of five experienced endoscopists ranged from 0.449 to 0.531 in diagnosing GIST, and from 0.702 to 0.809 in distinguishing group 1 and 4 GIST. The overall performance of AORMS is better than that of experienced endoscopists. In the prospective validation set, the AORMS also achieved 0.770 and 0.750 AUC for the diagnosis and risk stratification of small GISTs (<2cm).

Table 1. Diagnostic performance of the AORMS.

	AUC	ACC	SENS	SPEC	PPV	NPV	LR+	LR-
Model development cohort
Diagnosis of GIST	0.762	0.708	0.819	0.649	0.750	0.644	2.333	0.279
Risk stratification of GIST	0.734	0.872	0.571	0.908	0.571	0.897	6.207	0.472
Independent validation cohort
Diagnosis of GIST	0.770	0.742	0.662	0.806	0.774	0.644	3.412	0.419
Risk stratification of GIST	0.750	0.872	0.667	0.757	0.667	0.795	2.745	0.440

AORMS, automatically optimized radiomics modeling system; ACC, accuracy; AUC, area under the curve; GIST, gastrointestinal stromal tumor; LR+, positive likelihood ratio; LR-, negative likelihood ratio; NPV, negative predictive value; PPV, positive predictive value; SENS, sensitivity; SPEC, specificity.

Conclusion

AORMS showed good and stable performance in diagnosis and risk stratification of GISTs, which may help endoscopists provide more accurate diagnosis for small gastric SMTs (<2cm).

Conference

UEG Week Copenhagen 2023

Topics

Oesophagus

Submission format

Abstract

Session

Advanced endoscopy (Posters)

Citation

United European Gastroenterology Journal 2023; 11 (Supplement 8)

Published

2023

More Like This:

EFFECT OF SAME-SESSION ENDOSCOPIC ULTRASONOGRAPHY ON ENDOSCOPIC RETROGRADE CHOLANGIOPANCREATOGRAPHY IN PANCREATICOBILIARY DISEASES: A RANDOMIZED CONTROLLED TRIAL

Zefeng Xing

Multidisciplinary clinics in liver cirrhosis: The new digital era in decompensated cirrhosis

Mattias Mandorfer

Words of welcome

Maura Corsetti

THE HIDDEN IMPACT OF CELIAC DISEASE ON FEMALE REPRODUCTIVE HEALTH

Alessandro Repici

Beyond surgery: The expanding role of bariatric and metabolic endoscopy

Ivo Boskoski

UNDERWATER ENDOSCOPIC MUCOSAL RESECTION OF RECTAL NEUROENDOCRINE TUMORS – A CASE SERIES OF AN INNOVATIVE ENDOSCOPIC APPROACH

Guilherme Macedo

UEG Presentation

New

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

EFFECT OF SAME-SESSION ENDOSCOPIC ULTRASONOGRAPHY ON ENDOSCOPIC RETROGRADE CHOLANGIOPANCREATOGRAPHY IN PANCREATICOBILIARY DISEASES: A RANDOMIZED CONTROLLED TRIAL

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.