Related Experiment Video
Updated: Jun 24, 2026

Establishment and Evaluation of a Risk Prediction Model for Pathological Escalation of Gastric Low-Grade Intraepithelial Neoplasia
Published on: February 16, 2024
EGID: A Comprehensive Multi-label Endoscopic Image Dataset for Gastritis Classification
Zeyu Li1, Yingying Fang1, Yiming Song1
1Division of Gastroenterology and Hepatology, Key Laboratory of Gastroenterology and Hepatology, Ministry of Health, Shanghai Institute of Digestive Disease, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China.
A new dataset of endoscopic images aids in classifying chronic atrophic gastritis, a precancerous condition. This resource supports AI development for better gastritis diagnosis and cancer risk assessment.
Area of Science:
- Gastroenterology
- Medical Imaging
- Artificial Intelligence
Background:
- Chronic atrophic gastritis is a precancerous condition requiring accurate endoscopic classification.
- Existing endoscopic datasets lack comprehensive multi-dimensional annotations for systematic gastritis classification.
Purpose of the Study:
- To introduce the Endoscopic Gastritis Image Dataset (EGID), the first publicly available multi-label endoscopic gastritis dataset.
- To enable the development of AI-based classification systems for gastritis.
- To serve as an educational resource for clinical training in gastritis diagnosis and cancer risk stratification.
Main Methods:
- Collected 5,883 white-light endoscopic images from 229 patients.
- Annotated images across four dimensions: H. pylori infection status, presence of gastric atrophy, distribution of atrophy, and gastritis type.
- Ensured high inter-rater agreement (Cohen's Kappa: 0.947-0.970) for image-based annotations through independent assessment and consensus adjudication.
Main Results:
- The Endoscopic Gastritis Image Dataset (EGID) is now publicly available.
- The dataset features high-quality, multi-dimensionally annotated endoscopic images.
- Demonstrated excellent inter-rater reliability for endoscopic assessments.
Conclusions:
- EGID is the first multi-label endoscopic gastritis dataset, addressing a critical gap in existing resources.
- The dataset will facilitate the advancement of AI in gastritis classification and diagnosis.
- EGID will support clinical training and improve stratification of cancer risk in patients with gastritis.
Related Concept Videos
Peptic Ulcer Disease III: Clinical Manifestations and Diagnostic Studies
Few clinical manifestations differentiate gastric ulcers from duodenal ulcers. Distinctions in the location, timing, and pain relief are crucial for healthcare providers in differentiating between gastric and duodenal ulcers during clinical assessments.
Endoscopic Procedures I: Esophagogastroduodenoscopy
During an EGD, the endoscope can be used to:
Gastritis-I: Introduction and Types
Acute gastritis presents as a sudden inflammation triggered by various stressors to the stomach lining, such as exposure to corrosive agents, local irritants like aspirin and other NSAIDs, alcohol consumption, radiation therapy, physical trauma, severe burns, sepsis,...
Gastritis III: Clinical Manifestations and Management
Clinical manifestations of acute gastritis
The patient with acute gastritis may have a rapid onset of symptoms, such as epigastric pain or discomfort, dyspepsia, anorexia, hiccups, or nausea and vomiting, which can last from a few hours to a few days. Erosive or hemorrhagic gastritis may cause bleeding, which may manifest as blood in vomit or as...
Endoscopic Procedures III: Video Capsule Endoscopy
Gastritis-II: Pathophysiology
In acute gastritis, the gastric mucosa becomes swollen and red and undergoes superficial erosion. Superficial ulceration may lead to bleeding.
In chronic gastritis, persistent or repeated insults lead to chronic inflammatory changes and, eventually, thinning or atrophy of the gastric tissue.
Gastritis can stem from various causes, each...