Laboratory for Laser Molecular Imaging and Machine Learning

Scientific results:

An approach based on machine learning and spectral taxonomy methods, super-resolution reconstruction methods, deep neural networks to content analysis of the spectra of multicomponent samples based on spectral patterns and a priori information on the spectral characteristics of individual components, and the development of predictive models for subsequent classification was developed. Methods for screening non-invasive diagnostics of viral and bacterial respiratory infections based on spectral analysis:

• exhaled air using laser optical-acoustic spectroscopy and AI methods;
• biological fluids using laser Raman spectroscopy and AI methods were developed. A provocative test for assessing the presence of viral respiratory infections on the mucous membrane of the oral cavity and bronchopulmonary system using laser spectroscopy was developed.

Implementation of research results:

Negotiations are underway with potential investors on the implementation of medical diagnostics technology based on spectral analysis of exhaled air samples using laser optical-acoustic spectroscopy and machine learning.

Organizational and infrastructural changes:

As a result of the project implementation, a laboratory (analytical center) was created, equipped with scientific equipment for laser spectroscopy, specialized software implementing machine and deep learning in the field of creating predictive models for classifying laser spectroscopy data. For this purpose, the original laser gas analyzer was modernized, combining optical-acoustic detection and a laser source based on parametric light generation. A complex for Raman spectroscopy in the terahertz and IR regions was created, including in the mode of giant and locally enhanced Raman scattering spectroscopy.

Education and personnel occupational retraining:

In 2021, two programs of the discipline were developed and implemented into the educational process as part of the implementation of the master's program:

• "Medical diagnostics using machine learning and biophotonics".
• "Introduction to bioinformatics".

In 2022, two programs of the discipline were developed and implemented into the educational process as part of the implementation of the master's program:

• "Mathematical models of biophotonics methods".
• "Visualization methods in biology and medicine".

In 2022, the Candidate Exam Program in the scientific specialty 1.3.21 Medical Physics was developed and implemented into the educational process.

In 2023, three programs of the discipline were developed and implemented into the educational process as part of the implementation of the master's program:

• "Mathematical models of biophotonics methods".
• "Visualization methods in biology and medicine".
• "Neurophotonics".

As a result of the project implementation, 5 dissertations for the degree of candidate of physical and mathematical sciences, 6 master's dissertations, and 3 bachelor's degrees were defended.

The laboratory staff, under the guidance of the leading scientist, completed 2 scientific internships: at the Lomonosov Moscow State University on the topic of "Analysis of THz spectra of exhaled air using machine learning methods" (Russia, Moscow), at the Institute of Laser and Information Technology Problems of the Russian Academy of Sciences on the topic of "Creation of an optical-acoustic detector based on a quartz resonator" (Russia, Shatura).

Cooperation:

• University at Albany (SUNY, USA).
• University of Lorraine (France).
• Lomonosov Moscow State University.
• Saratov National Research University named after N.G. Chernyshevsky