Nuclei Segmentation in Cervical Cytology Images Using Pix2Pix and U-Net

• João Pedro Lobato de Pinho PUC Minas
• Alexei Manso Correa Machado PUC Minas / UFMG

DOI: https://doi.org/10.5753/sbcas.2025.6952

Abstract

The segmentation of cell nuclei is essential for the analysis of cytological images and the early detection of diseases such as cervical cancer. This work compares the U-Net and Pix2Pix architectures in the task of semantic segmentation of cervical cell nuclei. Pix2Pix is a conditional GAN that uses adversarial learning. The experiments were conducted using the CNSeg dataset, which contains cytological images in various contexts. U-Net achieved better performance in AJI (+2.90%) and PQ (+3.12%) metrics. However, Pix2Pix showed promising results by producing accurate segmentations. These findings suggest that adversarial networks can be a promising alternative for this task, which remains underexplored in the literature.

References

Ali, M., Ali, M., Hussain, M., and Koundal, D. (2024). Generative adversarial networks (gans) for medical image processing: Recent advancements. ARCH COMPUT METHOD E.

Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. C., and Bengio, Y. (2014). Generative adversarial networks. COMMUN ACM, 63:139 – 144.

Isola, P., Zhu, J.-Y., Zhou, T., and Efros, A. A. (2017). Image-to-image translation with conditional adversarial networks. PROC CVPR IEEE, pages 5967–5976.

Khaled, A., Han, J.-J., and Ghaleb, T. A. (2022). Multi-model medical image segmentation using multi-stage generative adversarial networks. IEEE Access, 10:28590–28599.

Luc, P., Couprie, C., Chintala, S., and Verbeek, J. (2016). Semantic segmentation using adversarial networks. ArXiv, abs/1611.08408.

Mahmood, F., Borders, D., Chen, R. J., Mckay, G. N., Salimian, K. J., Baras, A., and Durr, N. J. (2020). Deep adversarial training for multi-organ nuclei segmentation in histopathology images. IEEE T MED IMAGING, 39(11):3257–3267.

Mirza, M. and Osindero, S. (2014). Conditional generative adversarial nets. ArXiv, abs/1411.1784.

Park, K.-B., Choi, S. H., and Lee, J. Y. (2020). M-gan: Retinal blood vessel segmentation by balancing losses through stacked deep fully convolutional networks. IEEE Access, 8:146308–146322.

Popescu, D., Deaconu, M., Ichim, L., and Stamatescu, G. (2021). Retinal blood vessel segmentation using pix2pix gan. MED C CONTR AUTOMAT, pages 1173–1178.

Ronneberger, O., Fischer, P., and Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. ArXiv, abs/1505.04597.

Siegel, R. L., Giaquinto, A. N., and Jemal, A. (2024). Cancer statistics, 2024. CA: A Cancer Journal for Clinicians, 74(1):12–49.

Son, J., Park, S. J., and Jung, K.-H. (2019). Towards accurate segmentation of retinal vessels and the optic disc in fundoscopic images with generative adversarial networks. J DIGIT IMAGING, 32(3):499–512.

van der Schot, A., Sikkel, E., Niekolaas, M., Spaanderman, M., and de Jong, G. (2023). Placental vessel segmentation using pix2pix compared to u-net. Journal of Imaging, 9(10).

Xue, Y., Xu, T., Zhang, H., Long, L. R., and Huang, X. (2018). Segan: Adversarial network with multi-scale l1 loss for medical image segmentation. Neuroinformatics, 16(3):383–392.

Zhang, X., Zhu, X., Zhang, X., Zhang, N., Li, P., and Wang, L. (2018). Seggan: Semantic segmentation with generative adversarial network. 2018 IEEE Fourth International Conference on Multimedia Big Data (BigMM), pages 1–5.

Zhao, J., jun He, Y., Zhou, S.-H., Qin, J., and ning Xie, Y. (2023). Cnseg: A dataset for cervical nuclear segmentation. COMPUT METH PROG BIO, 241:107732.