RU 
EN 
CN 
ES 
Studies by the scientists of Centre for Artificial Intelligence at Samara National Research University headed by Dr. Yury Kovsh, researcher of the Department of Radiology and Biomedical Imaging at Yale School of Medicine, will help improve quality and accuracy of measurements in functional MRI (fMRI) brain scanning of patients with a variety of psychoneurological disorders and those undergoing post-apoplexy neurorehabilitation.
Findings of the academic research and experiments will be presented as a dedicated module and used for upgrading OpenNFT.org* software platform widely used by a number of the world’s leading universities as well as research and development centers to visualize and analyze activity of the brain scanned in real time.
Research is held under the grant by the Russian Foundation for Basic Research (RFBR). The grant is provided for a two-year project in a total amount of RUB 1.2 million. The first findings have been summarized in the paper "Recurrent quality control of functional MRI supported by OpenNFT" presented in the collection of studies "Information technology and nanotechnology".**
"In our project, we aim to investigate and improve real-time brain fMRI quality parameters, specifically, in the system for generating neurobiological feedback. Building of such feedback on the basis of fMRI enables rehabilitation of patients with psychoneurological disorders and deviations, e.g. schizophrenia, autism, posttraumatic stress, hyperactivity, attention deficit etc. Clearly, this system is also used in scientific research of brain involving healthy volunteers", - says Nikita Davydov, teaching assistant in the Samara University Department of Engineering Cybernetics.
According to him, statistically, about 50% data obtained in pediatric research through brain fMRI are unreliable and need to be excluded from further analysis.
"We expect our solution to reduce the share of unreliable data due to obtaining a more complete picture of quality parameters, enabling timely correction of distortions during scanning, and thereby, reducing the cost of scanning, diagnosis and rehabilitation on the basis of neurointerface, and increasing procedure accuracy", - he explained.
From practical experience, most of the scanning data distortions occur due to head movements.
"For instance, a person can become nervous in a confined space, start moving and even panicking. Precursors of such behavior need to be recognized at an early stage, and the experiment or therapy has to be suspended or even discontinued. This is especially relevant in examination of children. If a child starts moving, our module lets the operator know in advance that the scanning session has to be interrupted or repeated in a while", - Nikita Davydov points out.
The neural network will process a series of quality parameters to effectively reflect various quality aspects of the scanning process. Samara researchers obtain data for experimental validation of the technology under development from Yale University and public domain databases.
"Our solutions will be available for free use in scope of OpenNFT.org project, both in research purposes and for medical applications, to enable more effective scanning, diagnosis and rehabilitation based on neurointerface. Scanning quality parameters’ tracking and control in real time will help detect minor anomalies in the data and reduce human subjectivity in research and clinical applications", - Nikita Davydov concludes.
For reference
* OpenNFT.org software platform has been developed with participation of scientists Centre for Artificial Intelligence at Samara University and Image Processing Systems Institute of the RAS as part of the international research group of scientists headed by Yu.А. Kovsch, including scientists of the University College of London, École polytechnique fédérale de Lausanne and University of Zurich. OpenNFT.org is intended for conducting research related to generation of neurobiological feedback on the basis of MRI scanner. In the course of experiments, the person in the scanner receives information about his or her own brain activity e.g. in the form of visual or audial feedback. The software suite also visualizes brain activity and quality parameters of scanned images for the operator.
OpenNFT.org is an open platform whose operating principles are specified in the work: Y.Koush et al., OpenNFT: An open-source Python/Matlab framework for real-time fMRI neurofeedback training based on activity, connectivity and multivariate pattern analysis, Neuroimage, 2017.
** Davydov N.S., Prilepin Е., Auer T., Gninenko N., Khramov A.G., Van de Ville D., Nikonorov A.V., Kovsh Yu.А. Recurrent quality control of functional MRI supported by OpenNFT. //information technology and nanotechnology (ITNT-2020). P.192-197.
