MEDICAL APPLICATIONS OF IR AND THZ LASER MOLECULAR IMAGING AND MACHINE LEARNING Текст научной статьи по специальности «Медицинские технологии»

MEDICAL APPLICATIONS OF IR AND THZ LASER MOLECULAR IMAGING

AND MACHINE LEARNING

YU.V.KISTENEV1'3, A.V.BORISOV A.I.KNYAZKOVA1,2, D.A.VRAZHNOV1,2, V.E SKIBA1,

V.V. PRISCHEPA1, A.CUISSET4

1National Research Tomsk State University, Tomsk, Russia 2Institute of Strength Physics and Materials Science SB RAS, Tomsk, Russia 3Siberian State Medical University, Tomsk, Russia

4Université du Littoral Côte d'Opale, Dunkerque, France

Abstract

The problem of relevant information extracting from 2D and 3D laser molecular imaging experimental data, which can be used for medical diagnosis, is of great importance. Due to the high dimension of molecular imaging data, methods of mathematical statistics become inefficient. To overcome this problem, we use a machine learning approach. A typical machine learning pipeline is shown in Fig. 1.

r

Data Feature selection/ extraction Model training Model validation => Predictions

t

Figure 1: A typical machine learning pipeline. The peculiarities of molecular imaging in IR and THz spectral ranges, methods of extracting of informative features from experimental data, and creating of prognostic models for medical diagnosis using machine learning methods will be discussed for three fields of applications: diagnostics using breath air or biofluids spectral analysis, and noninvasive tissue spectroscopy and visualization.

This work was performed under the Government statement of work for ISPMS Project No. III.23.2.10, and with partial financial support from the Russian Foundation for Basic Research, Grant No. №17-0000275 (17-00-00186).