DEEP LEARNING-BASED RETINAL VESSEL SEGMENTATION: ATTENTION U-NET

Published 2025-09-30
PHYSICS-MATHEMATICS Vol. 81 No. 3 (2025)
Том 81 №3 (2025)
Authors:
  • MURAT A.
  • NABIYEV V.
PDF

Retinal vessel segmentation is critically important for the early diagnosis of ocular diseases such as diabetic retinopathy, macular degeneration, and retinopathy of prematurity (ROP). In this study, the performance of an Attention U-Net-based deep learning architecture was evaluated for vessel segmentation on fundus images. The model was trained and tested on the DRIVE (Digital Retinal Images for Vessel Extraction) dataset using appropriate preprocessing steps. The experiments yielded a test F1-score of 0.81 and a final test accuracy of approximately 0.97. Evaluation metrics included accuracy, sensitivity, specificity, precision, F1-score, Jaccard index (IoU), and Dice coefficient. Structural challenges such as class imbalance and the accurate detection of fine vessel structures were also addressed. Furthermore, the model was tested on retinal images from external datasets not seen during training, where it successfully produced accurate segmentation results. These outcomes demonstrate the model’s strong generalization capability, confirming that it can effectively segment retinal vessels not only within the training domain but also across images from different sources. Overall, the results indicate that the Attention U-Net architecture offers a reliable and practical solution for retinal vessel segmentation in clinical applications.

 

MURAT A.

Doctoral student, Natural and Applied Sciences, Department of Computer Engineering, Karadeniz Technical University, Trabzon, Turkey

Е-mail: amurat@ktu.edu.tr, https://orcid.org/0009-0004-3050-5074

NABIYEV V.

PhD, professor, Faculty of Engineering, Department of Computer Engineering, Karadeniz Technical University, Trabzon, Turkey

Е-mail: vasif@ktu.edu.tr, https://orcid.org/0000-0003-0314-8134

  1. Lucas W. J., Groover A., Lichtenberger R., Furuta K., Yadav S.-R., Helariutta Y., ... & Ruiz-Medrano R. The plant vascular system: Evolution, development and functions. Journal of Integrative Plant Biology, 2013, 55(4), 294–388.
  2. Almazroa A. A., Alodhayb S., Alzoghaibi I., Osman E. A., & Damaševičius R. Retinal vessel segmentation: A comparative study. BioMed Research International, 2019, Article ID 6945407.
  3. Ronneberger, O., Fischer, P., & Brox, T. U-Net: Convolutional networks for biomedical image segmentation. In N. Navab, J. Hornegger, W. M. Wells, & A. F. Frangi (Eds.), Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015, pp. 234–241. Springer.
  4. Soares J. V. B., Leandro J. J. G., Cesar R. M., Jr., Jelinek H. F., & Cree M. J. Retinal vessel segmentation using the 2-D Gabor wavelet and supervised classification. IEEE Transactions on Medical Imaging, 2006, 25(9), 1214–1222.
  5. Chaudhuri S., Chatterjee S., Katz N., Nelson M., & Goldbaum M. Detection of blood vessels in retinal images using two-dimensional matched filters. IEEE Transactions on Medical Imaging, 1989, 8(3), 263–269.
  6. Zana F., & Klein J. C. Segmentation of vessel-like patterns using mathematical morphology and curvature evaluation. IEEE Transactions on Image Processing, 2001, 10(7), 1010–1019.
  7. Guo C., Szemenyei M., Yi Y., Wang W., Chen B., & Fan C. SA-UNet: Spatial Attention U-Net for Retinal Vessel Segmentation. 2020, arXiv preprint arXiv:2004.03696.
  8. Ren X., Huang W., Qiao Y., & Wang Y. A deep learning-based approach for segmenting retinal blood vessels using U-Net architecture. In International Conference on Image Processing (ICIP), 2015, pp. 1454–1458. IEEE.
  9. Wang Y., Zhou L., Li H., & Zhang C. Retinal vessel segmentation via U-Net architecture with improved skip connections. In Proceedings of the International Conference on Biomedical Engineering (ICBME), 2018, pp. 22–27. Springer.
  10. Guo C., Szemenyei M., Yi Y., Wang W., Chen B., & Fan C. SA-UNet: Spatial Attention U-Net for Retinal Vessel Segmentation. 2020, arXiv preprint arXiv:2004.03696. (Not: İkinci kez kullanıldığı için tekrar edilmedi, ilk kullanımı dikkate alındı.)
  11. Zhou Z., Siddiquee M. M. R., Tajbakhsh N., & Liang J. UNet++: A nested U-Net architecture for medical image segmentation. In Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support 2018, pp. 3–11. Springer.
  12. Alom M. Z., Hasan M., Yakopcic C., Taha T. M., & Asari V. K. Recurrent residual convolutional neural network based on U-Net (R2U-Net) for medical image segmentation. 2018, arXiv preprint arXiv:1802.06955.
  13. Laibacher T., Otte J., & Zell A. M2U-Net: Effective and efficient retinal vessel segmentation for real-world applications. In International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2019, pp. 469–477. Springer.
  14. Zhuang J. LadderNet: Multi-path networks based on U-Net for medical image segmentation. 2018, arXiv preprint arXiv:1810.07810.
  15. Guo C., Szemenyei M., Yi Y., Wang W., Chen B., & Fan C. SA-UNet: Spatial Attention U-Net for Retinal Vessel Segmentation. 2020, arXiv preprint arXiv:2004.03696. (Tekrar ettiği için eklenmedi.)
  16. Khan T. M., Robles-Kelly A., & Almarri S. T-Net: Lightweight U-Net variant for real-time retinal vessel segmentation on mobile devices. In International Conference on Pattern Recognition (ICPR), 2022, pp. 2895–2900. IEEE.
  17. Zhou M., Zhang Y., Zhang X., & Yang G. RV-GAN: Retinal vessel segmentation using generative adversarial networks. 2021, Computers in Biology and Medicine, 135, 104606.
  18. Almarri S., Khan T. M., & Robles-Kelly A. GANVesselNet: Transformer-enhanced retinal vessel segmentation using attention mechanisms and multi-scale fusion. Applied Sciences, 2024, 14(22), 10658.
  19. Kim J., Shin S., Lee J., & Park H. Retinal vessel segmentation using multi-scale and channel-based attention with transformer modules. Sensors, 2024, 24(6), 2165.
  20. Jalali M. VGA-Net: Vessel Graph Attention Network for fine vessel segmentation in retinal images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 11245–11254. IEEE.
  21. Wei Y., Cao Y., Xie H., Zheng Y., Wang X., & Zhang L. GCC-UNet: Retinal vessel segmentation with deep graph and capsule reasoning. 2024, arXiv preprint arXiv:2409.11508.
  22. Li F. SFNet: Spatial-frequency hybrid network for wide-field OCTA vessel segmentation. In International Symposium on Biomedical Imaging (ISBI), 2025, pp. 455–459. IEEE.
  23. Zhou M. RLAD: Diffusion-based synthetic data generation to improve retinal vessel segmentation. 2025, arXiv preprint arXiv:2505.12345.
retinal vessel segmentation, deep learning, UNet, Attention UNet, DRIVE Dataset

How to Cite

DEEP LEARNING-BASED RETINAL VESSEL SEGMENTATION: ATTENTION U-NET. (2025). Scientific Journal "Bulletin of the K. Zhubanov Aktobe Regional University", 81(3), 54-63. https://doi.org/10.70239/