NTC Module Modulmed

Main articles:

• Neuronets (neural networks)
• Artificial intelligence in medicine
• Artificial intelligence in radiology

2021: Representation Modulmed

On February 1, 2021 the Module Scientific and technological center, the representative in area of hardware solutions in the field of artificial intelligence and microelectronic design, provided the development – the software Modulmed for recognition of viral pneumonia, including, the caused COVID-19 infection. Given software of Module it is capable to recognize a disease with an accuracy of 93 percent.

Implementing work of neural networks of deep learning, software Modulmed allows to reveal in several minutes pathology of lungs in an analysis result of radiological pictures of patients. It considerably saves time of highly skilled medical staff which is forced to work with large volumes of clinical information. The software creates three-dimensional visualization of pictures, allows to determine the volume, area and extent of damage of lungs. For recognition of affected areas of lungs the deep neural network with architecture of U-Net is used. The network studied and tested on databases of the Federal medical biophysical center of A.I. Burnazyan. Specialists of this center also carried out work on a marking of the images received by methods of a computer tomography.

This software is easily integrated into the existing IT infrastructure medical the organizations. In the long term we are going to expand considerably its functionality that will allow to apply software not only in pulmonology, but also in other spheres of medicine and also to diagnose more wide range of diseases, - Alexey Mokhnatkin, the associate director on development of STC Module comments.

For February, 2021 software Modulmed is a part of the demonstrator of the neural network hardware and software system (NNHSS). NPAK is the scalable computing communication framework intended for creation and implementation in practice of products of medical AI. The solution created by specialists of STC Module in cooperation with the Resource center of universal design and rehabilitation technologies" (FGAU "RTsUD and RT") consists of the interconnected automated jobs (AJ) in which domestic NeuroMatrix microprocessors are used. Automated workplaces are integrated by a local network with data center which has necessary computing resources for storage of large volumes of data and deep learning of artificial neural networks. In the long term the complex will become the replicated and entrusted solution and will allow to solve a big range of relevant problems in the field of domestic medicine. First of all is a creation of an ecosystem for development and deployment of services of medical AI and ensuring access of doctors to these services, implementation of domestic software and hardware tools in medical practice and also integration with the existing digital medical infrastructure.