| Radiology Room |
| Ultrasound Room |
| Surgery Room |
| Laboratory Room |
| Comprehensive Room |
| Pediatrics Room |
| Dental Room |
| Medical operation instruments |
| Hospital Furniture |
| Medical supplies |
News Center
AI Detects Stomach Cancer Risk from Upper Endoscopic Images
In many parts of the world, doctors must make complex clinical decisions with limited access to specialist support, advanced diagnostics, or pathology services. This is especially challenging in gastrointestinal care, where early signs of stomach disease can be subtle and easily missed during routine endoscopy. As a result, conditions such as Helicobacter pylori (H. pylori) infection and early precancerous changes may go undetected. Researchers have now developed an artificial intelligence (AI) system that can assist frontline physicians by interpreting standard endoscopy images with expert-level insight.
The multi-model AI framework developed by researchers at National Taiwan University Hospital (Taipei, Taiwan) is designed to replicate how experienced endoscopists and pathologists assess gastric health. By learning from large datasets of endoscopy images matched with pathology-confirmed diagnoses, the system captures expert reasoning that is typically acquired only after many years of clinical practice.
The AI platform consists of several interconnected models that work sequentially. It first selects high-quality endoscopic images, then identifies anatomically appropriate regions of the stomach, and finally analyzes surface and vascular patterns associated with disease. Training incorporated both visual features and histopathological ground truth, allowing the system to link what is seen endoscopically with underlying tissue changes.
Using routine endoscopy images, the AI system accurately identified signs of H. pylori infection and early gastric mucosal changes associated with increased cancer risk. The system provided standardized and detailed assessments, reducing reliance on non-specific descriptors such as “gastritis.” The results, published in Endoscopy, demonstrate that AI can extract clinically meaningful information from images already collected in daily practice.
By offering fast, consistent image interpretation, the AI tool can support physicians working in low-resource or remote settings. It may help guide decisions on H. pylori eradication therapy, the need for additional diagnostic testing, or follow-up surveillance. More broadly, the system has the potential to reduce disparities in care quality between well-resourced medical centers and underserved regions by embedding expert-level analysis directly into routine workflows.
“AI is not meant to replace doctors,” said corresponding author Professor Yi-Chia Lee. “It acts as a digital assistant that supports clinical judgment. By fitting into routine care, AI helps bring more consistent medical quality to reduce the gap between well-resourced hospitals and remote communities.”
http://www.allmycbs.com/en/index.asp






