ISSN Print 2500–1094
ISSN Online 2542–1204
BIOMEDICAL JOURNAL OF PIROGOV UNIVERSITY (MOSCOW, RUSSIA)
Copyright: © 2025 by the authors. Licensee: Pirogov University.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (CC BY).
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (CC BY).
ORIGINAL RESEARCH
Development of the vascular condition classifier using supervised machine learning methods
About authors
1 Vladimir Zelman Center for Neurobiology and Neurorehabilitation Skolkovo Institute of Science and Technology, Moscow
2 Artificial Intelligence Center Skolkovo Institute of Science and Technology, Moscow
3 Center for Photonics and Photonic Technologies Skolkovo Institute of Science and Technology, Moscow
4 Center for Molecular and Cellular Biology Skolkovo Institute of Science and Technology, Moscow
Correspondence should be addressed: Zlata Besedovskaia
Bolshoy Bulvar, 30, Building 1, Moscow, 121205; moc.liamg@dlogantari
About paper
Acknowledgments: All authors of this article express their gratitude to the authors of article [15] for providing the open data used in this study.
Author contribution: Z. Besedovskaya — development of the pipeline and clustering tools, image preparation for publication, and draft publication. A. Korobov — creation and integration of new vessel segment features into the pipeline and draft publication. N. Kudryashova — medical conceptualization and validation of the vessel segment features and draft publication. All authors contributed equally to this study.
Received: 2025-10-07
Accepted: 2025-11-10
Published online: 2025-11-27
|
Fig. 1. A. Schematic representation of the experimental design. B. Schematic representation of the experimental setup. C. Image in the analyzed coordinate plane. Modified from [15]
Fig. 2. UMAP clustering of the original data
Fig. 3. Characteristics of trained ML models: Catboost, SGDClassifier, LogisticRegression. A) Feature weights for individual vessel segments. B) Error matrices for the three models
Table 1. Developed and calculated vessel features to describe the characteristics of vascular segments and their position in the image area
Table 2. Obtained metrics of unsupervised clustering
Table 3. Metrics for assessing the quality of supervised classification models
