EGID: 5,883 Multi-label Endoscopic Gastritis Images from Shanghai, 2024

5,883 high-quality white-light endoscopic images from 229 patients were retrospectively collected at Renji Hospital, Shanghai, between March and November 2024. The dataset, created by Zeyu Li, provides systematic multi-label annotations across four clinical dimensions for gastritis classification. Annotations were validated by experienced endoscopists, achieving excellent inter-rater agreement.

Use Cases

• Training multi-label classification models for H. pylori infection status based on endoscopic images.
• Developing AI systems to detect and classify gastric atrophy and its distribution patterns.
• Creating educational tools for clinical training in gastritis diagnosis and cancer risk stratification.
• Benchmarking model performance on a dataset with validated, expert-level annotations.

Strengths

• 5,883 images from 229 patients provide a substantial sample for model development.
• Annotations validated by two endoscopists with a senior reviewer, achieving a Cohen's Kappa of 0.947-1.000.
• Systematic labeling across four independent clinical dimensions (H. pylori status, atrophy presence, atrophy distribution, gastritis type).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for specific modeling tasks.
• Data is from a single hospital (Renji Hospital, Shanghai) over a limited time range, which may reflect geographic and temporal bias.

Provenance

Source

Retrospectively collected at Renji Hospital, Shanghai, using the Olympus CV-290 system.

Collection Method

Clinical image collection with expert annotation and consensus review.

Time Range

March to November 2024

Freshness

Last updated 2026-05-04 15:47:37; freshness should be verified.

Geography

Shanghai, China