informИнформатикаInformatics1816-03012617-6963UIIP NASB10.37661/1816-0301-2026-23-2-21-38inform-1405Research ArticleОБРАБОТКА СИГНАЛОВ, ИЗОБРАЖЕНИЙ, РЕЧИ, ТЕКСТА И РАСПОЗНАВАНИЕ ОБРАЗОВSIGNAL, IMAGE, SPEECH, TEXT PROCESSING AND PATTERN RECOGNITIONПредсказание центров ядер раковых клеток на иммуногистохимических флуоресцентных изображенияхPrediction of cancer cell nuclear centers in immunohistochemical fluorescence imageshttps://orcid.org/0000-0003-0880-4188СкакунВ. В.SkakunV. V.

Скакун Виктор Васильевич, кандидат физико-математических наук, доцент, заведующий кафедрой системного анализа и компьютерного моделирования

пр. Независимости, 4, Минск, 220030

Victor V. Skakun, Cand. Sci. (Phys.-Math.), Assoc. Prof., Head of Department System Analysis and Computer Modeling

av. Nezavisimosti, 4, Minsk, 220030

skakun@bsu.byhttps://orcid.org/0000-0002-1356-0897СюйС.XuS.

Сюй Сылунь, соискатель кафедры системного анализа и компьютерного моделирования

пр. Независимости, 4, Минск, 220030

Silun Xu, Applicant at the Department of System Analysis and Computer Modeling

av. Nezavisimosti, 4, Minsk, 220030

xusilun@hotmail.comБелорусский государственный университетBelarusian State University2026280620262322138Copyright © Скакун В.В., Сюй С., 20262026Скакун В.В., Сюй С.Skakun V.V., Xu S.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://inf.grid.by/jour/article/view/1405Цели

Цели. Цель исследования состоит в разработке метода предсказания центров ядер раковых клеток на иммуногистохимических флуоресцентных изображениях с использованием точечной разметки центров ядер.

Методы

Методы. В работе используются сверточные нейронные сети глубокого обучения.

Результаты

Результаты. Предложен метод предсказания центров ядер раковых клеток на иммуногистохимических флуоресцентных изображениях пораженных тканей, отличающийся от существующих использованием точечной разметки центров ядер при обучении сети. Разработан алгоритм предварительной и заключительной обработки изображений, позволяющий проводить полный цикл анализа изображений любой размерности.

Заключение

Заключение. Метод предсказания центров ядер раковых клеток на иммуногистохимических флуоресцентных изображениях имеет простую архитектуру, небольшое количество обучаемых параметров и не требует сложной постобработки результатов анализа, традиционной при семантической сегментации ядер клеток, заключающейся в разделении слипшихся ядер. Метод позволяет подсчитывать количество раковых клеток на единицу площади, что в свою очередь предоставляет возможность оценить степень заболевания. Полное время анализа изображения размером 2048×2048 пикселей с использованием вычислителя T4 (Google Colab) составляет в среднем 750 мс, что открывает возможности анализа полнослайдовых изображений высокой размерности за приемлемое время.

Objectives

Objectives. The aim of the study is to develop a method for predicting cancer cell nuclear centers in immunohistochemical fluorescence images using point annotation of nuclear centers.

Methods

Methods. Deep learning convolutional neural networks are used in this study.

Results

Results. A method for predicting cancer cell nuclear centers in immunohistochemical fluorescence images of diseased tissues is proposed. The method differs from existing approaches by using point annotation of nuclear centers during the learning process. An algorithm for image pre- and postprocessing has been developed, enabling an end-to-end analysis for images of any dimension.

Conclusion

Conclusion. A method for predicting cancer cell nuclear centers in immunohistochemical fluorescence images has been developed. It has a simple architecture, a small number of trainable parameters, and does not require complex post-processing of analysis results, traditionally involved in semantic segmentation for separating clustered nuclei. The method allows for counting the number of cancer cells per unit area, which in turn makes it possible to assess the extent of the disease. The total analysis time for a 2048×2048 pixel image using the T4 (Google Colab) compute engine averages 750 ms, enabling the analysis of high-dimensional, whole-slide images in a reasonable time.

иммуногистохимические изображенияизображения раковых клетокпредсказание центров ядерподсчет клетокнейронные сети глубокого обученияархитектура U-Netimmunohistochemical imagingcancer cell imagingnuclear center predictioncell countingdeep learning neural networksU-Net architectureReferencesCamp, R. L. Automated subcellular localization and quantification of protein expression in tissue microarrays / R. L. Camp, G. G. Chung, D. L. Rimm // Nature Medicine. – 2002. – Vol. 8, no. 11. – P. 1323–1327. – https://doi.org/10.1038/NM791.Camp, R. L. Automated subcellular localization and quantification of protein expression in tissue microarrays / R. L. Camp, G. G. Chung, D. L. Rimm // Nature Medicine – 2002. V. 8, № 11. – P. 1323-1327. DOI:10.1038/NM791Quantitative analysis of estrogen receptor heterogeneity in breast cancer / G. G. Chung, M. P. Zerkowski, S. Ghosh [et al.] // Laboratory Investigation. – 2007. – Vol. 87, no. 7. – P. 662–669. – https://doi.org/10.1038/LABINVEST.3700543.Quantitative analysis of estrogen receptor heterogeneity in breast cancer / G.G. Chung, M.P. Zerkowski, S. Ghosh, R.L. Camp and D.L. Rimm // Lab Invest. – 2007. – V. 87, № 7. – P. 662-669. DOI:10.1038/LABINVEST.3700543A review and comparison of breast tumor cell nuclei segmentation performances using deep convolutional neural networks / A. Lagree, M. Mohebpour, N. Meti [et al.] // Scientific Reports. – 2021. – Vol. 11, no. 1. – URL: https://www.nature.com/articles/s41598-021-87496-1 (date of access: 21.01.2026). – https://doi.org/10.1038/S41598-021-87496-1.A review and comparison of breast tumor cell nuclei segmentation performances using deep convolutional neural networks / A. Lagree, M. Mohebpour, N. Meti, K. Saednia, F.I. Lu, E. Slodkowska, S. Gandhi, E. Rakovitch, A. Shenfield, A. Sadeghi-Naini, W.T. Tran // – Sci. Rep. – 2021. – V. 11, № 1: 8025. DOI:10.1038/S41598-021-87496-1Automating cell counting in fluorescent microscopy through deep learning with c-ResUnet / R. Morelli, L. Clissa, R. Amici [et al.] // Scientific Reports. – 2021. – Vol. 11, no. 1. – URL: https://www.nature.com/articles/s41598-021-01929-5 (date of access: 21.01.2026). – https://doi.org/10.1038/s41598-021-01929-5.Automating cell counting in fluorescent microscopy through deep learning with c-ResUnet / R. Morelli, L. Clissa, R. Amici, M. Cerri, T. Hitrec, M. Luppi, L. Rinaldi, F. Squarcio, A. Zoccoli // Sci. Rep. – 2021. V. 11, № 1: 2920. DOI:10.1038/s41598-021-01929-5Srinidhi, C. L. Deep neural network models for computational histopathology: A survey / C. L. Srinidhi, O. Ciga, A. L. Martel // Medical Image Analysis. – 2021. – Vol. 67. – Art. 101813. – URL: https://www.sciencedirect.com/science/article/abs/pii/S1361841520301778?via%3Dihub (date of access: 21.01.2026). – https://doi.org/10.1016/j.media.2020.101813.Srinidhi, C. L. Deep neural network models for computational histopathology: A survey / C. L. Srinidhi, O. Ciga, A. L. Martel // Med. Image Anal. – 2021. – V. 67: 101813. DOI:10.1016/j.media.2020.101813.Xie, W. Microscopy cell counting and detection with fully convolutional regression networks / W. Xie, J. A. Noble, A. Zisserman // Computer Methods in Biomechanics and Biomedical Engineering. – 2018. – Vol. 6, no. 3. – P. 283–292. – https://doi.org/10.1080/21681163.2016.1149104.Xie, W. Microscopy cell counting and detection with fully convolutional regression networks / W. Xie, J. A. Noble, A. Zisserman // Comput. Methods Biomech. – 2018. – V. 6, № 3. – P. 283–292. DOI:10.1080/21681163.2016.1149104Сюй, С. Сравнительный анализ нейронных сетей глубокого обучения для сегментации ядер клеток на иммуногистохимических флуоресцентных изображениях раковых тканей / С. Сюй, В. В. Скакун // Журнал Белорусского государственного университета. Математика. Информатика. – 2024. – № 1. – С. 59–70.Xu, Silun. Comparative analysis of deep learning neural networks for the segmentation of cancer cell nuclei on immunohistochemical fluorescent images / Silun Xu, VV Skakun // Journal of the Belarusian State University. Mathematics and Informatics. – 2024. – V. 1– P. 59 –70. Russian.Ronneberger, O. U-Net: Convolutional networks for biomedical image segmentation / O. Ronneberger, P. Fischer, T. Brox // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 : 18th Intern. Conf., Munich, Germany, 5–9 Oct. 2015. – Munich, 2015. – Part III. – P. 234–241. – https://doi.org/10.1007/978-3-319-24574-4_28.Ronneberger, O. U-Net: Convolutional Networks for Biomedical Image Segmentation / O. Ronneberger, P. Fischer, T. Brox // MICCAI. – 2015. P. 234-241. DOI:10.1007/978-3-319-24574-4_28.UNet 3+: A full-scale connected UNet for medical image segmentation / H. Huang, L. Lin, R. Tong [et al.] // IEEE Intern. Conf. on Acoustics, Speech and Signal Processing (ICASSP), Barcelona, Spain, 04–08 May 2020. – Barcelona, 2020. – P. 1055–1059. – https://doi.org/10.1109/ICASSP40776.2020.9053405.UNet 3+: A Full-Scale Connected UNet for Medical Image Segmentation / H. Huang, L. Lin, R. Tong, H. Hu, Q. Zhang, Y. Iwamoto, X. Han, Y. W. Chen and J. Wu // IEEE International Conference on Acoustics, Speech and Signal Processing. – 2020. – P. 1055–1059. DOI:10.1109/ICASSP40776.2020.9053405.TransUNet: Transformers make strong encoders for medical image segmentation / J. Chen, Y. Lu, Q. Yu [et al.]. – 2021. – URL: https://arxiv.org/pdf/2102.04306 (date of access: 21.01.2026). – https://doi.org/10.48550/arXiv.2102.04306.TransUNet: Transformers Make Strong Encoders for Medical Image Segmentation / J. Chen, Y. Lu, Q. Yu, X. Luo, E. Adeli, Y. Wang, L. Lu, A. L. Yuille, Y. Zhou // arXiv. – 2021. – 2102.04306v1. DOI:10.48550/arXiv.2102.04306.Li, W. U-ResNet, a novel network fusion method for image classification and segmentation / W. Li, Z. Gao, Y. Song // Sensors. – 2025. – Vol. 25, no. 17. – Art. 5600. – URL: https://www.mdpi.com/1424-8220/25/17/5600 (date of access: 21.01.2026). – https://doi.org/10.3390/S25175600.Li, W. U-ResNet, a Novel Network Fusion Method for Image Classification and Segmentation / W. Li, Z. Gao, Y. Song // Sensors. – 2025. – V. 25, № 17: 5600. DOI:10.3390/S25175600Mask R-CNN / K. He, G. Gkioxari, P. Dollár, R. Girshick // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2017. – Vol. 42, no. 2. – P. 386–397. – https://doi.org/10.1109/TPAMI.2018.2844175.Mask R-CNN / K. He, G. Gkioxari, P. Dollár, R. Girshick // IEEE Trans. Pattern Anal. Mach. Intell. – 2017. – V. 42, № 2. – P. 386–397. DOI:10.1109/TPAMI.2018.2844175Faster R-CNN: Towards real-time object detection with region proposal networks / S. Ren, K. He, R. Girshick, J. Sun // IEEE Transactions on Pattern Analysis and Machine Intelligence. – 2015. – Vol. 39, no. 6. – P. 1137–1149. – https://doi.org/10.1109/TPAMI.2018.2844175.Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks / S. Ren, K. He, R. Girshick, J. Sun // IEEE Trans. Pattern Anal. Mach. Intell. – 2015. – V. 39, № 6. – P. 1137–1149. DOI:10.1109/TPAMI.2018.2844175Жоголь, Д. П. Анализ иммуногистохимических флуоресцентных изображений с помощью нейронных сетей глубокого обучения / Д. П. Жоголь, Л. Лисица, В. В. Скакун // Компьютерные технологии и анализ данных (CTDA’2024) : материалы IV Междунар. науч.-практ. конф., Минск, 25–26 апр. 2024 г. / Белорус. гос. ун-т ; редкол.: В. В. Скакун (гл. ред.), Н. Н. Яцков, В. В. Гринёв. – Мн. : БГУ, 2024. – C. 15–18.Zhogol, D. P. Analysis of immunohistochemical fluorescence images using deep learning neural networks / D. P. Zhogol, L. Lisitsa, V. V. Skakun // Computer technologies and data analysis (CTDA’2024) [Electronic resource]: Proc. IV Int. sci.-pract. conf., Minsk, April 25–26, 2024 / Belarusian state University; editorial board: V. V. Skakun (editor-in-chief), N. N. Yatskov, V. V. Grinev. – Minsk: BSU, 2024. – P. 15–18.Count-ception: Counting by fully convolutional redundant counting / J. P. Cohen, G. Boucher, C. A. Glastonbury [et al.] // IEEE Intern. Conf. on Computer Vision Workshops (ICCV Workshops), Venice, Italy, 22–29 Oct. 2017. – Venice, 2017. – P. 18–26. – https://doi.org/10.1109/ICCVW.2017.9.Count-ception: Counting by Fully Convolutional Redundant Counting / J. P. Cohen, G. Boucher, C. A. Glastonbury, H. Z. Lo, Y. Bengio // IEEE International Conference on Computer Vision Workshops. – 2017. – P. 18–26. DOI:10.1109/ICCVW.2017.9Deep learning and transfer learning for automatic cell counting in microscope images of human cancer cell lines / F. Lavitt, D. J. Rijlaarsdam, D. van der Linden [et al.] // Applied Sciences. – 2021. – Vol. 11, no. 11. – Art. 4912. – URL: https://www.mdpi.com/2076-3417/11/11/4912 (date of access: 21.01.2026). – https://doi.org/10.3390/APP11114912.Deep Learning and Transfer Learning for Automatic Cell Counting in Microscope Images of Human Cancer Cell Lines / F. Lavitt, D. J. Rijlaarsdam, D. van der Linden, E. Weglarz-Tomczak and J. M. Tomczak // Applied Sciences. – 2021. – V. 11, № 11: 4912. DOI:10.3390/APP11114912Beyond classification: Structured regression for robust cell detection using convolutional neural network / Y. Xie, F. Xing, X. Kong [et al.] // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 : 18th Intern. Conf., Munich, Germany, 5–9 Oct. 2015. – Munich, 2015. – Part III. – P. 358–365. – https://doi.org/10.1007/978-3-319-24574-4_43.Beyond Classification: Structured Regression for Robust Cell Detection Using Convolutional Neural Network / Y. Xie, F. Xing, X. Kong, X. Su, L. Yang // Lecture Notes in Computer Science, Springer. – 2015. – V. 9351. – P. 358–365. DOI:10.1007/978-3-319-24574-4_43Deep voting: A robust approach toward nucleus localization in microscopy images / Y. Xie, X. Kong, F. Xing [et al.] // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015 : 18th Intern. Conf., Munich, Germany, 5–9 Oct. 2015. – Munich, 2015. – Part III. – P. 374–382. – https://doi.org/10.1007/978-3-319-24574-4_45.Deep Voting: A Robust Approach Toward Nucleus Localization in Microscopy Images / Y. Xie, X. Kong, F. Xing, F. Liu, H. Su, L. Yang // Med. Image Comput. Comput. Assist. Interv. – 2015. – V. 9351. – P. 374–382. DOI:10.1007/978-3-319-24574-4_45Mitosis detection in breast cancer histology images with deep neural networks / D. C. Ciresan, A. Giusti, L. M. Gambardella, J. Schmidhuber // Medical Image Computing and Computer-Assisted Intervention – MICCAI 2013 : 16th Intern. Conf., Nagoya, Japan, 22–26 Sept. 2013. – Nagoya, 2013. – Part II. – P. 411–418. – https://doi.org/10.1609/aaai.v30i1.10140.Mitosis Detection in Breast Cancer Histology Images with Deep Neural Networks / D. C. Ciresan, A. Giusti, L. M. Gambardella, J. Schmidhuber // MICCAI. – 2013. – Part II, LNCS 8150. – P 411–418. DOI:10.1609/aaai.v30i1.10140Jin, X. CellRegNet: Point annotation-based cell detection in histopathological images via density map regression / X. Jin, H. An, M. Chi // Bioengineering. – 2024. – Vol. 11, no. 8. – Art. 814. – URL: https://www.mdpi.com/2306-5354/11/8/814 (date of access: 21.01.2026). – https://doi.org/10.3390/BIOENGINEERING11080814.Jin, X. CellRegNet: Point Annotation-Based Cell Detection in Histopathological Images via Density Map Regression / X. Jin, H. An, M. Chi // Bioengineering. – 2024. – V. 11, № 8: 814. DOI:10.3390/BIOENGINEERING11080814Proximity adjusted centroid mapping for accurate detection of nuclei in dense 3D cell systems / T. Van De Looverbosch, S. De Beuckeleer, F. De Smet [et al.] // Computers in Biology and Medicine. – 2025. – Vol. 185. – Art. 109561. – URL: https://www.sciencedirect.com/science/article/pii/S0010482524016469?via%3Dihub (date of access: 21.01.2026). – https://doi.org/10.1016/j.compbiomed.2024.109561.Van De Looverbosch T. и др. Proximity adjusted centroid mapping for accurate detection of nuclei in dense 3D cell systems/ T. V. De Looverbosch, S. De Beuckeleer, F. De Smet, J. Sijbers, W. H. De Vos // Computers in Biology and Medicine. – 2025. – V. 185: 109561. DOI: 10.1016/j.compbiomed.2024.109561Locality sensitive deep learning for detection and classification of nuclei in routine colon cancer histology images / K. Sirinukunwattana, S. E. Ahmed Raza, Y.-W. Tsang [et al.] // IEEE Transactions on Medical Imaging. – 2016. – Vol. 35, no. 5. – P. 1196–1206. – https://doi.org/10.1109/TMI.2016.2525803.Locality Sensitive Deep Learning for Detection and Classification of Nuclei in Routine Colon Cancer Histology Images / K. Sirinukunwattana, S. E. Ahmed Raza, Y-W Tsang, D. R. J. Snead, I. A. Cree, N. M. Rajpoot // IEEE Trans. Med. Imaging. – 2016. – V. 35, № 5. – P. 1196–1206. DOI:10.1109/TMI.2016.2525803Efficient and robust cell detection: A structured regression approach / Y. Xie, F. Xing, X. Shi [et al.] // Medical Image Analysis. – 2018. – Vol. 44. – P. 245–254. – https://doi.org/10.1016/j.media.2017.07.003.Efficient and robust cell detection: A structured regression approach /Y. Xie, F. Xing, X. Shi, X. Kong, H. Su, L. Yang // Med. Image Anal. – 2018. – V. 44. – P. 245–254. DOI:10.1016/j.media.2017.07.003Unleashing the power of prompt-driven nucleus instance segmentation / Z. Shui, Y. Zhang, K. Yao, C. Zhu // 18th European Conf. on Computer Vision, ECCV 2024, Milan, Italy, 29 Sept. – 4 Oct. 2024. – Milan, 2024. – Vol. 15085. – Р. 288–304. – https://doi.org/10.1007/978-3-031-73383-3_17.Shui, Z. et al. Unleashing the Power of Prompt-Driven Nucleus Instance Segmentation / Z. Shui, Y. Zhang, K. Yao, C. Zhu // ECCV 2024. Lecture Notes in Computer Science, Springer, – 2025. – V. 15085. https://doi.org/10.1007/978-3-031-73383-3_17Xu, S. Comparison of deep learning preprocessing algorithms of nuclei segmentation on fluorescence immunohistology images of cancer cells / S. Xu, V. V. Skakun // Pattern Recognition and Information Processing (PRIP'2021) : Proc. of the 15th Intern. Conf., 21–24 Sept. 2021, Minsk, Belarus / ed. A. V. Tuzikov [et al.]. – Springer, 2022. – P. 166–177.Xu, S. Comparison of Deep Learning Preprocessing Algorithms of Nuclei Segmentation on Fluorescence Immunohistology Images of Cancer Cells / S. Xu, V. V. Skakun // Pattern Recognition and Information Processing. PRIP 2021 / ed. Tuzikov A. V. Springer. – 2022. – P. 166–177.

The authors declare that there are no conflicts of interest present.