- Статьи
- Science and technology
- Probability History: AI diagnoses precancerous gastritis with 96% accuracy%
Probability History: AI diagnoses precancerous gastritis with 96% accuracy%
Russia has developed a digital pathologist based on artificial intelligence that can speed up the diagnosis of atrophic gastritis, a dangerous precancerous disease of the stomach in which stomach cells change their structure. The algorithm determines the presence and stage of pathology with an accuracy of 96%, reducing the workload of doctors and helping to start treatment on time. At the same time, experts note that it is not worth relying solely on the program. For more information about the technology, see the Izvestia article.
AI will detect atrophic gastritis
A Russian digital pathologist created on the basis of artificial intelligence will accelerate the diagnosis of atrophic gastritis, a precancerous condition of the stomach, five times. He will help specialists in making a diagnosis and prescribing therapy to a patient in order to minimize the risk of developing cancer.
Atrophic gastritis is a chronic disease of the stomach in which its mucous membrane becomes thinner and changes its structure. Cells that are atypical for the stomach appear in it, the so—called goblet cells, or intestinal metaplasia cells that resemble intestinal cells. The disease can occur due to a variety of reasons - alcohol abuse, prolonged inflammation, infection, autoimmune processes, and so on. This condition is dangerous because it can degenerate into stomach cancer. Therefore, when symptoms of atrophic gastritis appear (nausea, heartburn, belching, decreased appetite), it is important to make a quick diagnosis.
— Early diagnosis allows you to monitor the patient, prescribe treatment and prevent the development of cancer. And if the tumor does appear, remove it at the earliest stage and save the patient's life," explained Tatiana Demura, director of the Institute of Clinical Morphology and Digital Pathology at Sechenov University.
To confirm atrophic gastritis, the patient is subjected to an endoscopy, microscopic fragments of the gastric mucosa are obtained, processed and examined in multiple magnification. Previously, a microscope was used for this purpose; now, in modern clinics, micro-preparations are scanned and examined on special digital platforms, which makes it possible to automate the work of the laboratory.
Specialists from Medtech AI (a joint venture between Beeline and Sechenov University) have created an algorithm that diagnoses atrophic gastritis and determines its stage with 96% accuracy.
— The remaining 4% is a very small margin of error. For comparison, the results may vary from 5 to 15% for different doctors," commented Ekaterina Antonova, head of the development and implementation of digital products for medicine at Beeline.
The system classifies samples, identifies areas with pathology, and performs the necessary calculations. As a result, the speed of research increases by an average of five times. The algorithm was trained over the course of a year on the basis of 5.5 thousand images, indicating all abnormal zones on them so that artificial intelligence could compare new histological scans with samples.
The prospects of AI for the diagnosis of dangerous gastritis
The solution looks promising, but it is important to understand that this is part of a large global trend. There are already many AI developments in the world for the analysis of histological preparations, including for the diagnosis of precancerous conditions and oncological diseases, Albert Rizvanov, head of the Personalized Medicine Center of Excellence at Kazan (Volga Region) Federal University, told Izvestia.
— Today, atrophic gastritis is diagnosed by gastroscopy with biopsy and subsequent histological examination of the gastric mucosa. This is an accurate but time-consuming method, highly dependent on the experience of the pathologist and the workload of the laboratory. The disease is quite common. According to various sources, its signs are detected in 10-30% of the adult population, especially in older age groups and in cases of Helicobacter pylori infection, which makes early diagnosis fundamentally important," the scientist noted.
Digital AI services allow you to automatically find areas of atrophy and intestinal metaplasia, objectively assess the stage of the process and significantly speed up obtaining a conclusion. At the same time, the key constraint for their widespread adoption remains the healthcare economy. The current standards of medical care do not include additional costs for AI diagnostics yet, the expert noted.
— Atrophic gastritis is a common pathology. It develops against the background of chronic inflammation in the stomach. In medical practice, there are no difficulties with its diagnosis, since this disease is visible to the naked eye during endoscopy. AI can speed up the verification process, reduce the burden on the doctor, while maintaining the quality of diagnosis," Konstantin Kulaev, head of the Endoscopy Department at the Chelyabinsk Regional Clinical Center for Oncology and Nuclear Medicine, commented on the development.
AI as a diagnostic tool
Artificial intelligence in medicine works at several levels of prevention. Firstly, it speeds up diagnosis, allowing in certain cases to identify cancerous conditions before the development of oncology, said Alexandra Vatyan, senior researcher at the ITMO Institute of AI, NTI expert on artificial intelligence. Secondly, it automates the routine work of specialists, reducing the human factor and increasing the accuracy of the analysis. Thirdly, AI can process large amounts of data and, in some cases, identify patterns that are inaccessible to human perception.
The expert assessed the 96% diagnostic accuracy as an outstanding result — the system is not only more accurate, but also more versatile than specialists — it works regardless of the pathologist's experience.
AI with such accuracy and an error of about 4% makes fewer mistakes than doctors, confirmed Anton Averyanov, CEO of the ST IT group of companies, TechNet NTI market expert.
— Of course, you should not rely only on the program, but after clinical trials, I think there will be no doubt about positive diagnoses — this will allow you to apply the necessary treatment to patients on time, — said the expert.
Given the shortage of personnel in medicine, it is difficult to overestimate the importance of creating solutions that can increase the detection rate of pathologies, added Evgeny Alkhanov, CEO of EVA LAB LLC, market expert at NTI Helsnet.
In 2026, the developers plan to apply for registration of a medical device.
Переведено сервисом «Яндекс Переводчик»
