GIM-ENDO: A Multimodal Endoscopic Image and Video Dataset for Gastric Intestinal Metaplasia Morphology and Pathology

Gastric intestinal metaplasia (GIM) is a histopathologically confirmed precursor to gastric adenocarcinoma. Early, accurate detection remains a clinical challenge, and AI-assisted endoscopy is a promising solution. We present GIM-ENDO, a publicly available multimodal dataset comprising 98 endoscopic still images and 39 video clips from 24 patients (22 GIM-positive, 2 normal controls), acquired with the Olympus EVIS X1 system under white-light endoscopy (WLE), narrow-band imaging (NBI), and magnifying NBI (M-NBI). All cases are histopathologically confirmed. Annotations cover six IEE endoscopic signs (LBC, MTB, WOS, TV pattern/Fusion, atrophy, MLE), GIM subtype (complete/incomplete), and OLGA/OLGIM staging where available. The dataset is publicly accessible at https://doi.org/10.5281/zenodo.20707267. For the latest updates, refer to https://databiox.com.