AI-powered digital pathologist developed in Russia to diagnose gastritis
The technology is designed to support diagnosis and treatment decisions, helping reduce the risk of gastric cancer.
In Russia, atrophic gastritis can now be detected using artificial intelligence. The technology makes it possible to speed up diagnosis severalfold and, as a result, prescribe therapy in a timely manner.
Atrophic gastritis is a chronic stomach disease in which the cells of the gastric lining change their structure. Because this condition can lead to stomach cancer, making a timely and accurate diagnosis is critical.
Under conventional practice, patients undergo endoscopy, and fragments of the gastric mucosa are examined under a microscope. While this method is accurate, it is labor-intensive and depends on many factors, including human judgment. Digital AI-based tools automate the search for areas of atrophy and intestinal metaplasia, provide an objective assessment of disease stage, and significantly accelerate the delivery of diagnostic conclusions.
High accuracy and faster analysis
The system, developed by MedTech AI, a joint venture between Beeline and Sechenov University, classifies tissue samples, highlights pathological areas, and performs the required calculations. As a result, examinations are carried out roughly five times faster.
Over the course of a year, the algorithm was trained on 5,500 images. All abnormal areas were marked on the images, allowing the AI to compare new histological scans with reference samples.
The algorithm identifies the presence and stage of pathology with up to 96% accuracy. Experts note that a 4% margin of error for an algorithm is relatively small. Ekaterina Antonova, head of digital product development and implementation for healthcare at Beeline, points out that results from different physicians can vary by 5% to 15%.
In the near term, the developers plan to submit an application to register the system as a certified medical device.
