Artificial intelligence (AI) technologies are rapidly developing, and their application in medicine is becoming increasingly relevant. In this context, the work of scientists at the Biomedical Image Analysis Laboratory of the Joint Institute for Informatics and Problems of the National Academy of Sciences of Belarus (UIIP NAS Belarus) is particularly noteworthy. They won the NAS Belarus "TOP-10" competition for their achievements in developing a new method of federated learning and generative AI models. This event is the subject of an article, "Lungs under the Supervision of Neural Networks," in the new newspaper "Navuka."

In an interview, Vasily  Kovalev, a leading researcher at the laboratory, noted that the project is just beginning. Among the objectives is the creation of multimodal neural network models capable of simultaneously considering different types of input data, as is the case with a generalist. "Based on research in this area, experimental software will be developed, which, of course, after appropriate testing, will have to undergo certification before becoming available to a wide range of users. And this will take several years," the scientist says.

The practical aspects of the work are handled by junior research fellow Dmitry Karpenko, who has conducted large-scale computational experiments on federated training of generative neural networks for medical imaging of various modalities.

Laboratory Head Eduard  Snezhko emphasized the importance of continuing the research and noted that the team is moving in the right direction, possessing high levels of professional expertise. "This is challenging work, because when the final result is still far away, it's necessary to explain why this is important," he concluded.

Read the full article in the Navuka newspaper.