AI algorithm detects pathology in fluorographic images in 30 seconds

The Innopolis University algorithm has become the best in Russia for the analysis of fluorographic images using AI. The tests involved 21 companies with 39 services. During the study, Innopolis University conducted 43,000 studies of images in medical institutions in Moscow.

An experiment on the use of innovative technologies in the field of computer vision for the analysis of medical images was carried out by specialists from the Moscow Social Development Complex, the Information Technology Department, together with the Moscow Diagnostic and Telemedicine Center and the Health Department.

The platform for the full automation of fluorographic studies of the Institute of Artificial Intelligence of a Russian IT university has become a leader in the use of innovative technologies in the field of computer vision. They were used to analyze medical images and further use them in the Moscow healthcare system.

The specialists of the Institute of Artificial Intelligence have developed the AI ​​RADIOLOGY platform. It analyzes medical x-ray images and identifies a variety of lung pathologies, including pneumonia, tuberculosis, neoplasms, pneumothorax, and others. The neural network was trained on medical images obtained from healthcare organizations of the Republic of Tatarstan. Such automatic image analysis systems reduce the burden on radiologists and reduce the number of missed pathologies.

The service works in the cloud of Innopolis University and can be integrated with any medical organization. The process of processing the image, issuing, and sending the results takes 30 seconds. The service identifies 96% of studies with pathologies, the team of the IT university is working to improve the accuracy of the algorithm. The service is connected to the ERIS EMIAS system in Moscow, thanks to which doctors have access to most medical organizations in the capital of Russia.

“The project for the recognition of chest pathologies using X-ray images is the first project of Innopolis University in the field of artificial intelligence technologies,” says Ramil Kuleev, director of the Institute of Artificial Intelligence. – For more than six years we have been engaged in research work, participated in international competitions, developed, increased competencies. The complexity of artificial intelligence lies in the competent solution of many tasks, in each of which the smallest details significantly affect the final result: the formation of a dataset, the development of algorithms, the organization of infrastructure for development and testing.”

“The most important component is a dataset or training sample, which should be balanced and as adequate as possible to the data on which the service will have to work in real conditions,” concludes Kuleev. – Thanks to close cooperation with medical organizations of the Republic of Tatarstan, we managed to create such a dataset. And, of course, people are the key factor. Our team are real professionals involved in the project and focused on results. Participation in such an experiment for us is an opportunity to get an objective assessment of the expert community and evaluate the capabilities of the service in real conditions with a large flow of research.”

Author: John Kessler
Graduated From the Massachusetts Institute of Technology. Previously, worked in various little-known media. Currently is an expert, editor and developer of Free News.
Function: Director
E-mail: except.freenews@gmail.com