STUDY OF APPLICATION OF MASK R-CNN AND SEGMENT ANYTHING MODEL (SAM) FOR INSTANCE SEGMENTATION OF MESENCHYMAL STEM CELLS IN MICROPHOTOGRAPHS

M. V. Solopov1, E. S. Chechekhina2, A. G. Popandopulo1, A. S. Kavelina1

About the authors

M. V. Solopov1, E. S. Chechekhina2, A. G. Popandopulo1, A. S. Kavelina1

Попандопуло А. Г. – д.мед.н, профессор, заведующий лабораторией клеточно - тканевого культивирования ФГБУ «ИНВХ им. В.К. Гусака» Минздрава России, 283045, Донецк, пр-кт. Ленинский, 47. Область научных интересов: регенеративная биомедицина, клеточные технологии, хирургия

Download PDF

UDC: 004.93
DOI: 10.24412/2413-7383-2025-2-37-21-29
Language: Russian

Abstract: The study compares the efficiency of Mask R-CNN and Segment Anything Model (SAM) for instance segmentation of mesenchymal stem cells (MSCs) in microphotographs. Results on two datasets (low and high cell density) demonstrate that Mask R-CNN, fine-tuned on target data, outperforms SAM in mAP (0.64 vs. 0.61 for low density; 0.31 vs. 0.26 for high density). Increased cell density and overlap reduce accuracy for both models. SAM’s limitations in using only bounding boxes highlight the need for domain-specific adaptation.

Keywords: instance segmentation, mesenchymal stem cells, morphological heterogeneity, cell density, automated cell culture analysis, biomedical imaging, model adaptation.

References

1. Jovic D. Yu Y. Wang D. et al. A Brief Overview of Global Trends in MSC-Based Cell Therapy. Stem Cell Reviews and Reports. 2022. Vol. 18. P. 1525–1545. DOI: 10.1007/s12015-022-10369-1.
2. Khalid A. A.-A. Rehab Y. A.-A. Update on the Use of Mesenchymal Stem Cells in the Treatment of Various Infectious Diseases Including COVID-19 Infection. Journal of Stem Cell Therapy and Transplantation. 2023. Vol. 7. P. 034–042. DOI: 10.29328/journal.jsctt.1001033.
3. Chung M.-J. Son J.-Y. Park S. et al. Mesenchymal Stem Cell and MicroRNA Therapy of Musculoskeletal Diseases. International Journal of Stem Cells. 2021. Vol. 14. № 2. P. 150–167. DOI: 10.15283/ijsc20167.
4. Vargas-Rodríguez P. Cuenca-Martagón A. Castillo-González J. et al. Novel Therapeutic Opportunities for Neurodegenerative Diseases with Mesenchymal Stem Cells: The Focus on Modulating the Blood-Brain Barrier. International Journal of Molecular Sciences. 2023. Vol. 24. № 18. P. 14117. DOI: 10.3390/ijms241814117.
5. Adnan N. Umer F. Malik S. Implementation of Transfer Learning for the Segmentation of Human Mesenchymal Stem Cells—A Validation Study. Tissue and Cell. 2023. Vol. 83. P. 102149. DOI: 10.1016/j.tice.2023.102149.
6. Solopov M. Chechekhina E. Kavelina A. et al. Comparative Study of Deep Transfer Learning Models for Semantic Segmentation of Human Mesenchymal Stem Cell Micrographs. International Journal of Molecular Sciences. 2025. Vol. 26. № 5. P. 2338. DOI: 10.3390/ijms26052338.
7. Imboden S. Liu X. Lee B.S. et al. Investigating Heterogeneities of Live Mesenchymal Stromal Cells Using AI-Based Label-Free Imaging. Scientific Reports. 2021. Vol. 11. P. 6728. DOI: 10.1038/s41598- 021-85905-z.
8. Hoffman J. Zheng S. Zhang H. et al. Image-Based Discrimination of the Early Stages of Mesenchymal Stem Cell Differentiation. Molecular Biology of the Cell. 2024. Vol. 35. P. ar103. DOI: 10.1091/mbc.E24-02-0095.
9. Kong Y. Ao J. Chen Q. et al. Evaluating Differentiation Status of Mesenchymal Stem Cells by Label- Free Microscopy System and Machine Learning. Cells. 2023. Vol. 12. № 11. P. 1524. DOI: 10.3390/cells12111524.
10. Hassanlou L. Meshgini S. Alizadeh E. Evaluating Adipocyte Differentiation of Bone Marrow-Derived Mesenchymal Stem Cells by a Deep Learning Method for Automatic Lipid Droplet Counting. Computers in Biology and Medicine. 2019. Vol. 112. P. 103365. DOI: 10.1016/j.compbiomed.2019.103365.
11. Lan Y. Huang N. Fu Y. et al. Morphology-Based Deep Learning Approach for Predicting Osteogenic Differentiation. Frontiers in Bioengineering and Biotechnology. 2022. Vol. 9. P. 802794. DOI: 10.3389/fbioe.2021.802794. Исследование применения моделей Mask R-CNN и Segment Anything Model (SAM)… С
12. Mota S.M. Rogers R.E. Haskell A.W. et al. Automated Mesenchymal Stem Cell Segmentation and Machine Learning-Based Phenotype Classification Using Morphometric and Textural Analysis. Journal of Medical Imaging. 2021. Vol. 8. № 1. P. 014503. DOI: 10.1117/1.JMI.8.1.014503.
13. Ochs J. Biermann F. Piotrowski T. et al. Fully Automated Cultivation of Adipose-Derived Stem Cells in the StemCellDiscovery—A Robotic Laboratory for Small-Scale, High-Throughput Cell Production Including Deep Learning-Based Confluence Estimation. Processes. 2021. Vol. 9. № 4. P. 575. DOI: 10.3390/pr9040575.
14. Rettenberger L. Rieken Münke F. Bruch R. Reischl M. Mask R-CNN Outperforms U-Net in Instance Segmentation for Overlapping Cells. Current Directions in Biomedical Engineering. 2023. Vol. 9. № 1. P. 335–338. DOI: 10.1515/cdbme-2023-1084.
15. Murugesan R. T. G.A. A. V. et al. Instance Segmentation of Neuronal Cells Using U-Net, Mask R-CNN, and Detectron2. International Journal of Biomedical Engineering and Technology. 2024. Vol. 45. № 2. P. 129–149. DOI: 10.1504/IJBET.2024.138735.
16. Kirillov A. Mintun E. Ravi N. et al. Segment Anything. arXiv. 2023. [Электронный ресурс]. URL: https://arxiv.org/abs/2304.02643 (дата обращения: 01.07.2024).
17. Mazurowski M.A. Dong H. Gu H. et al. Segment Anything M odel for Medical Image Analysis: An Experimental Study. Medical Image Analysis. 2023. Vol. 89. P. 102918. DOI: 10.1016/j.media.2023.102918.
18. Ji W. Li J. Bi Q. et al. Segment Anything Is Not Always Perfect: An Investigation of SAM on Different Real-World Applications. Machine Intelligence Research. 2024. Vol. 21. P. 617–630. DOI: 10.1007/s11633-023-1385-0.
19. Huang Y. Yang X. Liu L. et al. Segment Anything Model for Medical Images? Medical Image Analysis. 2024. Vol. 92. P. 103061. DOI: 10.1016/j.media.2023.103061.
20. Ulman V. Maška M. Magnusson K.E.G. et al. An Objective Comparison of Cell-Tracking Algorithms. Nature Methods. 2017. Vol. 14. № 12. P. 1141–1152. DOI: 10.1038/nmeth.4473.

Issue: 2(37)'2025
Section: ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING
How to cite: M. V. Solopov1, E. S. Chechekhina2, A. G. Popandopulo1, A. S. Kavelina1. STUDY OF APPLICATION OF MASK R-CNN AND SEGMENT ANYTHING MODEL (SAM) FOR INSTANCE SEGMENTATION OF MESENCHYMAL STEM CELLS IN MICROPHOTOGRAPHS // Problems of Artificial Intelligence. - 2025. - № 2 (37). - P. 21-29. - https://paijournal.guiaidn.ru/en/2025/2(37)-3.html

Back to issue 2(37)'2025