We continue to offer our readers digests about the projects of Igor Sikorsky Kyiv Polytechnic Institute scientists, which won the grants of the National Research Foundation of Ukraine (the beginning in №29, September 24, 2020).
The outbreak of the coronavirus pandemic has become a global challenge for our planet. International institutions allocate significant funds to defeat the global danger. Countries, able to respond adequately to developments and offer effective and efficient approaches to minimize risks and threats to human health, will be able to gain additional political and economic benefits. Scientists from all countries and continents are rapidly developing appropriate methods and for diagnosis and treatment pharmaceuticals. Scientists of Igor Sikorsky Kyiv Polytechnic Institute are making a significant contribution to the fight against this year's plague. According to the results of the competition of the NRFU, the work of scientists from the Department of Computer Science "Artificial Intelligence Platform for Remote Automated Detection and Diagnosis of Human Diseases" should receive state funding.
At the request of the editorial board of the newspaper “Kyiv Politechnic” about the achievements of the research the team project manager, Head of the Department of Computer Science, Professor Sergii Stirenko said:
- Sergii Hryhorovych, how did this year's pandemic influence your developments?
- Since COVID-19 spreads exponentially, standard processing and decoding of images by computerized tomography (CAT scan), magnetic resonance imaging (MRI), and radiographic testing (X-ray) are difficult because it usually takes a physician 10-60 minutes to interpret each result. We offer the technology of analysis of images obtained by CAT, MRI and X-ray on the basis of artificial intelligence (AI) using deep neural networks, which can automatically help identify signs of 14 lung diseases (including coronavirus pneumonia) with numerical metrics of such probability. The advantages of the proposed AI system are high accuracy, which according to the tests of X-ray images for some lung diseases shows accuracy higher than the accuracy of radiologists (> 90%), and the processing duration of the X-ray image is less than 5 seconds. This is approximately 100-700 times faster than human processing and decoding, and therefore makes the procedure for early detection of signs of lung disease (screening) more effective, securing a high measuring accuracy at the same time.
- What image analysis technology is your system based on?
- The technology of the proposed AI system for the analysis of X-ray images is based on deep learning algorithms. As of September 1, 2020, the AI system has been tested for more than 200,000 X-ray images, which correspond to the ill-health of more than 30,000 people with 14 types of lung diseases.
The process of preparing the AI system for the analysis of X-ray images includes a learning phase and a prediction phase. During the training phase, differences between COVID-19 pneumonia, ordinary pneumonia, and other lung diseases or artifacts are studied. The AI system is able to predict the probability of COVID-19 pneumonia and ordinary pneumonia based on input X-ray images. This X-ray imaging technology can also mark areas of probable lung damage, helping physicians to measure disease progression effectively or treat patients with COVID-19.
- Who is a member of the author's team?
- The project team consists of: Senior Lecturer Oleg Alenin, Prof., Ph.D. of Physical and Mathematical Sciences Yuri Gordienko, Postgraduate Student Yuriy Kochura, Associate Professor, Ph.D. of Mathematics Mikhaylo Pavliuchenko, Associate Professor, Ph.D. of Technical Science Oleksandr Rokovy, Associate Professor, Ph.D. of Technical Sciences Anatolii Serhiienko, Prof., Ph.D. of Technical Sciences Sergiy Stirenko, Postgraduate Student Vlad Taran, Senior Lecturer Timur Shemsedinov.
And our partners: Shalimov National Institue of Surgery and Transplantation of the National Academy of Medical Sciences of Ukraine, the State Institution “National Research Center for Radiation Medicine of National Academy of Sciences of Ukraine” and the cloud provider “TANHOST” LLC.
- What can you say about the implementation of your project?
- To use of this AI system in practice, we propose to do the following:
1. Create a demo version of the AI system on the basis of Faculty of Informatics and Computer Science of Igor Sikorsky Kyiv Polytechnic Institute to get a remote demonstration of a complete medical report based on web service and cloud computing.
2. Create a full-scale medical report based on web service and cloud computing with the provision of X-ray-screening services based on global deployment and use of the AI system with remote access throughout Ukraine for pulmonologists and other relevant specialists, in order to screen risk groups proactively with shortages of the COVID-19 tests systems.
I would like to hope that the achievements of specialists of Igor Sikorsky Kyiv Polytechnic Institute will be worthily implemented and bring results.