Segmentation of the Glomerular Region in Pathological Kidney Slides

• Débora B. Nascimento UFPI
• Luciano Oliveira UFBA
• Washington Santos FIOCRUZ
• Angelo Duarte UEFS
• Kelson R. T. Aires UFPI
• Rodrigo M. S. Veras UFPI

DOI: https://doi.org/10.5753/eniac.2025.14444

Abstract

This study proposes a method for segmenting the glomerular region in renal histological images using convolutional neural networks (CNNs) based on U-Net and Sharp U-Net architectures with pretrained backbones. A total of 643 images stained with hematoxylin-eosin (HE), periodic acid-Schiff (PAS), and periodic acid-methenamine silver (PAMS) were evaluated, applying stratified 5-fold cross-validation. The U-Net with VGG-19 achieved the highest mean Dice score (95.45%), followed by the Sharp U-Net with DenseNet201. The results were consistent across staining techniques, with a slight advantage for PAMS and PAS. The method demonstrated accuracy and robustness, highlighting its potential as a diagnostic support tool in nephropathology.

References

Altini, N. et al. (2020). Semantic segmentation framework for glomeruli detection and classification in kidney histological sections. Electronics, 9(3):503.

Dimitri, G. M. et al. (2022). Deep learning approaches for the segmentation of glomeruli in kidney histopathological images. Mathematics, 10(11):1934.

Gadermayr, M. et al. (2019). Cnn cascades for segmenting sparse objects in gigapixel whole slide images. Comput. Med. Imaging Graph., 71:40–48.

Guyton, A. C. and Hall, J. E. (2017). Tratado de fisiologia médica. Elsevier, Rio de Janeiro, 13 edition.

Kohavi, R. (1995). A study of cross-validation and bootstrap for accuracy estimation and model selection. In Proceedings of the 14th International Joint Conference on Artificial Intelligence, volume 2, pages 1137–1143. Morgan Kaufmann Publishers Inc.

Litjens, G. et al. (2017). A survey on deep learning in medical image analysis. Medical Image Analysis, 42:60–88.

Oktay, O., Schlemper, J., Le Folgoc, L., Lee, M., Heinrich, M., Misawa, K., Mori, K., McDonagh, S., Hammerla, N. Y., Kainz, B., Glocker, B., and Rueckert, D. (2018). Attention u-net: Learning where to look for the pancreas. arXiv preprint arXiv:1804.03999.

Oliveira, D., Silva, A., Neves, L., Tosta, T., Martins, A., Faria, P., and Nascimento, M. (2025). Investigação de arquiteturas de aprendizagem profunda para segmentação de lesões em imagens histológicas. In Anais do XXV Simpósio Brasileiro de Computação Aplicada à Saúde, pages 617–628, Porto Alegre, RS, Brasil. SBC.

Ronneberger, O., Fischer, P., and Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 234–241. Springer.

Santos, J., Silva, R., Oliveira, L., Santos, W., Aldeman, N., Duarte, A., and Veras, R. (2024). Glomerulosclerosis detection with pre-trained cnns ensemble. Computational Statistics, 39:561–581.

Santos, J. D., Veras, R. M. d. S., and Silva, R. R. (2021). A hybrid of deep and textural features to differentiate glomerulosclerosis and minimal change disease from glomerulus biopsy images. Biomedical Signal Processing and Control, 70:103020.

Sociedade Brasileira De Nefrologia (2023). Glomerulopatias. Acesso em: 25 mar. 2025.

Tajbakhsh, N. et al. (2016). Convolutional neural networks for medical image analysis: full training or fine tuning? IEEE Transactions on Medical Imaging, 35(5):1299–1312.

Zhou, Z., Siddiquee, M. M. R., Tajbakhsh, N., and Liang, J. (2018). Unet++: A nested u-net architecture for medical image segmentation. In Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, pages 3–11. Springer.

Zhou, Z., Siddiquee, M. M. R., Tajbakhsh, N., and Liang, J. (2020). Unet++: Redesigning skip connections to exploit multiscale features in image segmentation. IEEE Transactions on Medical Imaging, 39(6):1856–1867.

Zunair, H. and Ben Hamza, A. (2021). Sharp u-net: Depthwise convolutional network for biomedical image segmentation. Computers in Biology and Medicine, 136:104699.