Tatyana A. Polevaya
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tatyana A. Polevaya has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Aleksandra Vatian
School of Translational Information Technologies, ITMO University, 197101 St. Petersburg, Russia
2 shared publications - Natalia Gusarova
ITMO University
2 shared publications - Alexander Golubev
ITMO University
1 shared publication - Alexey Zubanenko
ITMO UniversityITMO University
1 shared publication - Almaz G. Vanyurkin
Almazov National Medical Research Center
1 shared publication - Anna A. Tatarinova
Национальный медицинский исследовательский центр им. В.А. Алмазова 197341 Санкт-Петербург Российская Федерация
1 shared publication
Latest Publications
- Open Access
- 3 July 2025
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 25 (3) , 457-465
Deep learning-enhanced contour interpolation techniques for 3D carotid vessel wall segmentation
When studying human vessels using the contour interpolation method, there is a problem of insufficient data for training neural networks for automatic segmentation of the carotid artery wall. In this paper, automated methods of contour interpolation are proposed to expand the datasets, which allows for improved segmentation of vessel walls and atherosclerotic plaques. In this study, the performance of various interpolation methods is compared with the traditional nearest neighboring technique. A theoretical description and comparative evaluation of Linear, Polar, and Spline interpolation are presented. Quantitative metrics, including the Dice Similarity Coefficient, area and index differences, and normalized Hausdorff distances, are used to evaluate the performance of the methods. Performance evaluations are performed on various vessel morphologies for both the lumen and the outer wall boundaries. The study showed that Linear interpolation achieves better geometric performance (Cohen’s Kappa 0.92) and improved neural network performance (Score 0.86) compared to the State-of-the-Art model. The proposed interpolation methods consistently outperform nearest neighbor interpolation. Polar and spline methods are effective in generating anatomically plausible contours with improved smoothness and continuity, eliminating transition artifacts between slices. Statistical analysis confirmed good agreement and reduced variation of these methods. The results of the study are useful for the development of automated tools for assessing atherosclerotic plaque in carotid arteries, which is important for stroke prevention. Implementation of improved interpolation methods into clinical imaging workflows can significantly improve the reliability, accuracy, and clinical utility of vessel wall segmentation.
- 3 July 2025
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 25 (3) , 457-465
Open AccessDeep learning-enhanced contour interpolation techniques for 3D carotid vessel wall segmentation
When studying human vessels using the contour interpolation method, there is a problem of insufficient data for training neural networks for automatic segmentation of the carotid artery wall. In this paper, automated methods of contour interpolation are proposed to expand the datasets, which allows for improved segmentation of vessel walls and atherosclerotic plaques. In this study, the performance of various interpolation methods is compared with the traditional nearest neighboring technique. A theoretical description and comparative evaluation of Linear, Polar, and Spline interpolation are presented. Quantitative metrics, including the Dice Similarity Coefficient, area and index differences, and normalized Hausdorff distances, are used to evaluate the performance of the methods. Performance evaluations are performed on various vessel morphologies for both the lumen and the outer wall boundaries. The study showed that Linear interpolation achieves better geometric performance (Cohen’s Kappa 0.92) and improved neural network performance (Score 0.86) compared to the State-of-the-Art model. The proposed interpolation methods consistently outperform nearest neighbor interpolation. Polar and spline methods are effective in generating anatomically plausible contours with improved smoothness and continuity, eliminating transition artifacts between slices. Statistical analysis confirmed good agreement and reduced variation of these methods. The results of the study are useful for the development of automated tools for assessing atherosclerotic plaque in carotid arteries, which is important for stroke prevention. Implementation of improved interpolation methods into clinical imaging workflows can significantly improve the reliability, accuracy, and clinical utility of vessel wall segmentation.
- Open Access
- 1 December 2022
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 22 (6) , 1166
Automated evaluation of ECG parameters during the COVID-19 pandemic
Algorithms for prompt automated evaluation of electrocardiogram parameters in the absence of specialized equipment and specialized specialists are considered. The patient’s electrocardiogram is recorded on a paper tape, then it is photographed on the primary care doctor’s mobile phone and processed by a specialized application. The application digitizes the photographed image of the electrocardiogram, evaluates its main parameters as well as calculates criteria for the differential diagnosis of certain diseases using approximate formulas. In addition, the digitized electrocardiogram image is transmitted to the server and processed using a machine learning system. Algorithms for digitizing and analyzing an electrocardiogram have been developed that make it possible to evaluate its elements that are important for diagnosis, and the average error in determining the position of the most complex (smoothed) peaks — P and T waves — was no more than 0.1 mm. An algorithm for the criteria analysis of an electrocardiogram is proposed to support the differential diagnosis of acute myocardial infarction with ST segment elevation and early ventricular repolarization syndrome which provides accuracy values of 0.85 and F-scores of 0.74. An alternative algorithm based on a deep neural network is proposed which provides the best values — 0.96 and 0.88, respectively, but requires large computing resources and is executed on the server. The algorithms are implemented as a set of library functions. They can be used both independently and as part of a full-scale clinical decision support system for automated evaluation of electrocardiogram parameters based on a client-server architecture. In addition, all calculation results, together with a photograph of the original electrocardiogram, can be promptly transferred to a qualified cardiologist in order to receive an advisory opinion remotely.
- 1 December 2022
- Published by ITMO University in Scientific and Technical Journal of Information Technologies, Mechanics and Optics
- Vol. 22 (6) , 1166
Open AccessAutomated evaluation of ECG parameters during the COVID-19 pandemic
Algorithms for prompt automated evaluation of electrocardiogram parameters in the absence of specialized equipment and specialized specialists are considered. The patient’s electrocardiogram is recorded on a paper tape, then it is photographed on the primary care doctor’s mobile phone and processed by a specialized application. The application digitizes the photographed image of the electrocardiogram, evaluates its main parameters as well as calculates criteria for the differential diagnosis of certain diseases using approximate formulas. In addition, the digitized electrocardiogram image is transmitted to the server and processed using a machine learning system. Algorithms for digitizing and analyzing an electrocardiogram have been developed that make it possible to evaluate its elements that are important for diagnosis, and the average error in determining the position of the most complex (smoothed) peaks — P and T waves — was no more than 0.1 mm. An algorithm for the criteria analysis of an electrocardiogram is proposed to support the differential diagnosis of acute myocardial infarction with ST segment elevation and early ventricular repolarization syndrome which provides accuracy values of 0.85 and F-scores of 0.74. An alternative algorithm based on a deep neural network is proposed which provides the best values — 0.96 and 0.88, respectively, but requires large computing resources and is executed on the server. The algorithms are implemented as a set of library functions. They can be used both independently and as part of a full-scale clinical decision support system for automated evaluation of electrocardiogram parameters based on a client-server architecture. In addition, all calculation results, together with a photograph of the original electrocardiogram, can be promptly transferred to a qualified cardiologist in order to receive an advisory opinion remotely.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Preserving diversity in auxiliary objectives provably speeds up crossing plateaus
Using auxiliary objectives often speeds up optimization in different conditions. However, actual results depend heavily on how exactly these auxiliary objectives are used. In one of the methods which uses auxiliary objectives, the EA+RL method which uses reinforcement learning algorithms to choose between objectives, the following use case was considered. When the target objective has plateaus, reinforcement learning is unable to learn which auxiliary objectives can be used to cross these plateaus, and if an auxiliary objective is simply a refined version of the target one, reinforcement learning hardly helps, if at all. On the other hand, it is known that optimizing auxiliary objectives may help crossing plateaus, because it may introduce an additional slope which helps the optimizer to decide where to move. In this paper, we propose two ways to keep diversity in terms of auxiliary objectives while optimizing the target objective. We prove that simple duplicate elimination already speeds up optimization of XDIVK, a coarse-grained version of ONEMAX, when ONEMAX - or any injective function of unitation - is used as an auxiliary objective, provided the population size is large enough. When selection explicitly maximizes the spread of population according to ONEMAX, it is already fast when the population size is 2.
- 1 December 2016
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-8
Preserving diversity in auxiliary objectives provably speeds up crossing plateaus
Using auxiliary objectives often speeds up optimization in different conditions. However, actual results depend heavily on how exactly these auxiliary objectives are used. In one of the methods which uses auxiliary objectives, the EA+RL method which uses reinforcement learning algorithms to choose between objectives, the following use case was considered. When the target objective has plateaus, reinforcement learning is unable to learn which auxiliary objectives can be used to cross these plateaus, and if an auxiliary objective is simply a refined version of the target one, reinforcement learning hardly helps, if at all. On the other hand, it is known that optimizing auxiliary objectives may help crossing plateaus, because it may introduce an additional slope which helps the optimizer to decide where to move. In this paper, we propose two ways to keep diversity in terms of auxiliary objectives while optimizing the target objective. We prove that simple duplicate elimination already speeds up optimization of XDIVK, a coarse-grained version of ONEMAX, when ONEMAX - or any injective function of unitation - is used as an auxiliary objective, provided the population size is large enough. When selection explicitly maximizes the spread of population according to ONEMAX, it is already fast when the population size is 2.