<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Publishing DTD v1.3 20210610//EN" "JATS-journalpublishing1-3.dtd">
<article article-type="research-article" dtd-version="1.3" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xml:lang="ru"><front><journal-meta><journal-id journal-id-type="publisher-id">bsuir</journal-id><journal-title-group><journal-title xml:lang="ru">Доклады БГУИР</journal-title><trans-title-group xml:lang="en"><trans-title>Doklady BGUIR</trans-title></trans-title-group></journal-title-group><issn pub-type="ppub">1729-7648</issn><issn pub-type="epub">2708-0382</issn><publisher><publisher-name>БГУИР</publisher-name></publisher></journal-meta><article-meta><article-id pub-id-type="doi">10.35596/1729-7648-2025-23-6-103-109</article-id><article-id custom-type="elpub" pub-id-type="custom">bsuir-4254</article-id><article-categories><subj-group subj-group-type="heading"><subject>Research Article</subject></subj-group><subj-group subj-group-type="section-heading" xml:lang="ru"><subject>Статьи</subject></subj-group></article-categories><title-group><article-title>Многомасштабная нейронная сеть для классификации фрагментов гистологических изображений</article-title><trans-title-group xml:lang="en"><trans-title>Multi-Scale Neural Network for Classification of Histological Image Fragments</trans-title></trans-title-group></title-group><contrib-group><contrib contrib-type="author" corresp="yes"><name-alternatives><name name-style="eastern" xml:lang="ru"><surname>Косарева</surname><given-names>А. А.</given-names></name><name name-style="western" xml:lang="en"><surname>Kosareva</surname><given-names>A. A.</given-names></name></name-alternatives><bio xml:lang="ru"><p>Косарева Александра Андреевна, ассист. каф. электронной техники и технологии,; мл. науч. сотр.</p><p>220013, Минск, ул. П. Бровки, 6</p><p>Тел.: +375 17 293-88-60</p></bio><bio xml:lang="en"><p>Kosareva Aleksandra Andreevna, Assistant at the Electronic Engineering and Technology Department; Junior Researcher</p><p>220013, Minsk, P. Brovki St., 6</p><p>Tel.: +375 17 293-88-60</p></bio><email xlink:type="simple">kosareva@bsuir.by</email><xref ref-type="aff" rid="aff-1"/></contrib></contrib-group><aff-alternatives id="aff-1"><aff xml:lang="ru"><institution>Белорусский государственный университет информатики и радиоэлектроники; Объединенный институт проблем информатики Национальной академии наук Беларуси</institution></aff><aff xml:lang="en"><institution>Belarusian State University of Informatics and Radioelectronics; The United Institute of Informatics Problems of the National Academy of Sciences of Belarus</institution></aff></aff-alternatives><pub-date pub-type="collection"><year>2025</year></pub-date><pub-date pub-type="epub"><day>25</day><month>12</month><year>2025</year></pub-date><volume>23</volume><issue>6</issue><fpage>103</fpage><lpage>109</lpage><permissions><copyright-statement>Copyright &amp;#x00A9; Косарева А.А., 2025</copyright-statement><copyright-year>2025</copyright-year><copyright-holder xml:lang="ru">Косарева А.А.</copyright-holder><copyright-holder xml:lang="en">Kosareva A.A.</copyright-holder><license xml:lang="ru" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>Данная работа распространяется под лицензией Creative Commons Attribution 4.0.</license-p></license><license xml:lang="en" license-type="creative-commons-attribution" xlink:href="https://creativecommons.org/licenses/by/4.0/" xlink:type="simple"><license-p>This work is licensed under a Creative Commons Attribution 4.0 License.</license-p></license></permissions><self-uri xlink:href="https://doklady.bsuir.by/jour/article/view/4254">https://doklady.bsuir.by/jour/article/view/4254</self-uri><abstract><p>В статье представлена архитектура многомасштабной нейронной сети с несколькими входами, предназначенная для одновременной обработки групп фрагментов гистологических изображений, полученных при различных увеличениях. Предлагаемая модель интегрирует признаки на уровне скрытых слоев, что позволяет эффективно объединять информацию о тканевой структуре на разных масштабах детализации. Экспериментальные результаты показали, что применение трех уровней увеличения по одной стороне фрагмента (98,58; 197,16 и 394,32 мкм) обеспечивает оптимальный баланс между информативностью входных данных и устойчивостью модели. Использование архитектуры позволило повысить среднее значение F1-меры на 5 % по сравнению с одношкальным подходом, достигнув величины 0,8962 ± 0,0508, а в отдельных запусках – 0,9697. Наблюдаемое стандартное отклонение обусловлено не нестабильностью модели, а естественной вариабельностью медицинских данных в ходе формирования обучающих выборок. Полученные результаты подтверждают перспективность многомасштабного анализа для задач цифровой патологии и демонстрируют потенциал предложенного решения в качестве средства автоматизированного выделения подозрительных участков на полнослайдовых изображениях.</p></abstract><trans-abstract xml:lang="en"><p>This paper presents the architecture of a multi-scale, multi-input neural network for simultaneously processing groups of histological image fragments acquired at different magnifications. The proposed model integrates features at the hidden layer level, which allows for efficient fusion of tissue structure information at different detail scales. Experimental results have shown that applying three magnification levels on one side of a fragment (98.58; 197.16, and 394.32 μm) provides an optimal balance between the information content of the input data and the stability of the model. Using this architecture, the average F1-score increased by 5 % compared to a single-scale approach, reaching a value of 0.8962 ± 0.0508, and in some runs – 0.9697. The observed standard deviation is due to the natural variability of medical data during the formation of training samples, rather than to model instability. The obtained results confirm the promise of multiscale analysis for digital pathology tasks and demonstrate the potential of the proposed solution as a means of automated detection of suspicious areas in fullslide images.</p></trans-abstract><kwd-group xml:lang="ru"><kwd>многомасштабная нейронная сеть</kwd><kwd>глубокое обучение</kwd><kwd>гистопатологические изображения</kwd><kwd>полнослайдовые изображения</kwd><kwd>искусственный интеллект в медицине</kwd></kwd-group><kwd-group xml:lang="en"><kwd>multi-scale neural network</kwd><kwd>deep learning</kwd><kwd>histopathological images</kwd><kwd>full-slide images</kwd><kwd>artificial intelligence in medicine</kwd></kwd-group></article-meta></front><back><ref-list><title>References</title><ref id="cit1"><label>1</label><citation-alternatives><mixed-citation xml:lang="ru">Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection / L. Huang [et al.] // Frontiers in Neurorobotics. 2022. Vol. 16. DOI: 10.3389/fnbot.2022.881021.</mixed-citation><mixed-citation xml:lang="en">Huang L., Zhang Y., Yang X., Wang Z., Li H. (2022) Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection. Frontiers in Neurorobotics. 16. DOI: 10.3389/fnbot.2022.881021.</mixed-citation></citation-alternatives></ref><ref id="cit2"><label>2</label><citation-alternatives><mixed-citation xml:lang="ru">Liu Wenzhuo. Multi-Scale Unified Network for Image Classification / Liu Wenzhuo, Zhu Fei, Liu ChengLin // arXiv preprint. 2024. Mode of access: https://arxiv.org/abs/2405.12345. Date of access: 09.10.2025.</mixed-citation><mixed-citation xml:lang="en">Liu Wenzhuo, Zhu Fei, Liu Cheng-Lin (2024) Multi-Scale Unified Network for Image Classification. arXiv preprint. Available: https://arxiv.org/abs/2405.12345 (Accessed 9 October 2025).</mixed-citation></citation-alternatives></ref><ref id="cit3"><label>3</label><citation-alternatives><mixed-citation xml:lang="ru">Papillary Thyroid Carcinoma Semantic Segmentation Using Multi-Scale Adaptive Convolutional Network with Dual Decoders / T. Parasporins [et al.] // IEEE Access. 2025. Vol. 13. P. 17340–17353. DOI: 10.1109/ACCESS.2025.3532505.</mixed-citation><mixed-citation xml:lang="en">Parasporins T., Payatsuporn P., Jitaree A., Boonkong C. (2025) Papillary Thyroid Carcinoma Semantic Segmentation Using Multi-Scale Adaptive Convolutional Network with Dual Decoders. IEEE Access. 13, 17340–17353. DOI: 10.1109/ACCESS.2025.3532505.</mixed-citation></citation-alternatives></ref><ref id="cit4"><label>4</label><citation-alternatives><mixed-citation xml:lang="ru">Cell Detection in Pathology and Microscopy Images with Multi-Scale Fully Convolutional Neural Networks / X. Pan [et al.] // World Wide Web. 2018. Vol. 21. P. 1721–1743. DOI: 10.1007/s11280-017-0520-7.</mixed-citation><mixed-citation xml:lang="en">Pan X., Yang D., Li L., Xie W., Yang H., Chen S. (2018) Cell Detection in Pathology and Microscopy Images with Multi-Scale Fully Convolutional Neural Networks. World Wide Web. 21, 1721–1743. DOI: 10.1007/s11280-017-0520-7.</mixed-citation></citation-alternatives></ref><ref id="cit5"><label>5</label><citation-alternatives><mixed-citation xml:lang="ru">Методика формирования базы гистопатологических изображений папиллярного рака щитовидной железы для глубокого обучения / М. В. Фридман [и др.] // Информатика. 2023. Т. 20, №. 2. С. 28–38. DOI: 10.37661/1816-0301-2023-20-2-28-38.</mixed-citation><mixed-citation xml:lang="en">Fridman M. V., Kosareva A. A., Snezhko E. V., Kamlach P. V., Kovalev V. A. (2023) Papillary Thyroid Carcinoma Whole-Slide Images as a Basis for Deep Learning. Informatics. 20 (2), 28–38. DOI: 10.37661/1816-0301-2023-20-2-28-38 (in Russian).</mixed-citation></citation-alternatives></ref><ref id="cit6"><label>6</label><citation-alternatives><mixed-citation xml:lang="ru">Attention Is All You Need / A. Vaswani [et al.] // arXiv preprint. 2017. Mode of access: https://arxiv.org/abs/1706.03762. Date of access: 09.10.2025.</mixed-citation><mixed-citation xml:lang="en">Vaswani A., Shazeer N., Parmar N., Uszkoreit J., Jones L., Gomez A. N., et al. (2017) Attention Is All You Need. arXiv preprint. Available: https://arxiv.org/abs/1706.03762 (Accessed 9 October 2025).</mixed-citation></citation-alternatives></ref><ref id="cit7"><label>7</label><citation-alternatives><mixed-citation xml:lang="ru">Tan, М. EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks / M. Tan, Q. V. Le // Proceedings of the 36th International Conference on Machine Learning (ICML 2019), Long Beach, California, USA, 9–15 June 2019. Long Beach, 2019. P. 6105–6114.</mixed-citation><mixed-citation xml:lang="en">Tan M., Le Q. V. (2019) EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning (ICML 2019), Long Beach, California, USA, 9–15 June. Long Beach. 6105–6114.</mixed-citation></citation-alternatives></ref><ref id="cit8"><label>8</label><citation-alternatives><mixed-citation xml:lang="ru">Multi_input_network-Histologia [Electronic Resource] // GitHub. Mode of access: https://github.com/Ruiu4317/Multi_input_network-Histologia. Date of access: 14.10.2025.</mixed-citation><mixed-citation xml:lang="en">Multi_input_network-Histologia. GitHub. Available: https://github.com/Ruiu4317/Multi_input_networkHistologia (Accessed 14 October 2025).</mixed-citation></citation-alternatives></ref></ref-list><fn-group><fn fn-type="conflict"><p>The authors declare that there are no conflicts of interest present.</p></fn></fn-group></back></article>
