Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Apr 12;15(1):12549.
doi: 10.1038/s41598-025-90397-2.

Multiscale transformers and multi-attention mechanism networks for pathological nuclei segmentation

Yongzhao Du  1   2 , Xin Chen  3 , Yuqing Fu  3   4
Affiliations

Multiscale transformers and multi-attention mechanism networks for pathological nuclei segmentation

Yongzhao Du et al. Sci Rep. .

Abstract

Pathology nuclei segmentation is crucial of computer-aided diagnosis in pathology. However, due to the high density, complex backgrounds, and blurred cell boundaries, it makes pathology cell segmentation still a challenging problem. In this paper, we propose a network model for pathology image segmentation based on a multi-scale Transformer multi-attention mechanism. To solve the problem that the high density of cell nuclei and the complexity of the background make it difficult to extract features, a dense attention module is embedded in the encoder, which improves the learning of the target cell information to minimize target information loss; Additionally, to solve the problem of poor segmentation accuracy due to the blurred cell boundaries, the Multi-scale Transformer Attention module is embedded between encoder and decoder, improving the transfer of the boundary feature information and makes the segmented cell boundaries more accurate. Experimental results on MoNuSeg, GlaS and CoNSeP datasets demonstrate the network's superior accuracy.

PubMed Disclaimer

Conflict of interest statement

Declarations. Competing interests: The authors declare no competing interests.

Figures

Figure 1
Figure 1
The pathological nuclei segmentation framework with multiscale Transformers and multi-attention.
Figure 2
Figure 2
Channel-crossing transformer operation.
Figure 3
Figure 3
Channel-crossing attention operation.
Figure 4
Figure 4
Visualization of ablation experiment results. The first and second rows are the MoNuSeg dataset, the third and fourth rows are the GlaS dataset, and the fifth row is the CoNSeP dataset. (a) Original image, (b) Ground Truth, (c) Baseline(U-Net), (d) Baseline+Dense-CA, (e) Baseline+MSTA, (f) ours.
Figure 5
Figure 5
Qualitative comparisons. The first and third rows are the MoNuSeg dataset, the forth and sixth rows are the GlaS dataset, and the seventh and eighth row is the CoNSeP dataset. (a) Original image, (b) Ground Truth, (c) AttNet, (d) Swin-Unet, (e) UNet, (f) UCTransNet, (g) ours.
Figure 6
Figure 6
Detail comparison plot on the MoNuSeg dataset. (a) Original image, (b) Ground Truth, (c) AttNet, (d) Swin-Unet, (e) UNet, (f) UCTransNet, (g) ours.
Figure 7
Figure 7
FLOPs and total parameters for different models.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Juhong, A. et al. Super-resolution and segmentation deep learning for breast cancer histopathology image analysis. Biomed. Opt. Exp.14, 18–36 (2023). - PMC - PubMed
    1. Yu, K.-H. et al. Predicting non-small cell lung cancer prognosis by fully automated microscopic pathology image features. Nat. Commun.7, 12474 (2016). - PMC - PubMed
    1. Irshad, H., Veillard, A., Roux, L. & Racoceanu, D. Methods for nuclei detection, segmentation, and classification in digital histopathology: A review-current status and future potential. IEEE Rev. Biomed. Eng.7, 97–114 (2013). - PubMed
    1. Bian, X., Wang, G., Wu, Y., Li, Y. & Wang, H. Tci-unet: Transformer-cnn interactive module for medical image segmentation. Biomed. Opt. Exp.14, 5904–5920 (2023). - PMC - PubMed
    1. Liu, S. et al. A generative adversarial network based on deep supervision for anatomical and functional image fusion. Biomed. Signal Process. Control100, 107011 (2025).

LinkOut - more resources