数据与计算发展前沿 ›› 2026, Vol. 8 ›› Issue (4): 162-178.

CSTR: 32002.14.jfdc.CN10-1649/TP.2026.04.012

doi: 10.11871/jfdc.issn.2096-742X.2026.04.012

• • 上一篇    下一篇

PMDeUNet：一个用于高效分割腹部医学图像的轻量级网络模型

林予松1,2(),梁振羽1,赵哲()   

  1. 1 郑州大学网络空间安全学院， 河南 郑州 450002
    2 郑州大学互联网医疗与健康服务河南省协同创新中心， 河南 郑州 450052
  • 收稿日期:2025-11-04 出版日期:2026-08-20 发布日期:2026-08-21
  • 通讯作者: 赵哲（E-mail: sevenzz@zzu.edu.cn）
  • 作者简介:林予松，郑州大学网络空间安全学院，教授，互联网医疗与健康服务河南省协同创新中心，副主任。研究方向为深度学习、医学图像处理。
    本文中承担的工作：指导模型设计，编写模型代码。
    LIN Yusong is a professor at the School of Cyberspace Security, Zhengzhou University, and deputy director of the Henan Collaborative Innovation Center for Internet Medical and Health Services. His research interests include deep learning and medical image processing.
    In this paper, he is mainly responsible for guiding model design and writing the model code.
    E-mail: yslin@ha.edu.cn, 15039031716@163.com|赵哲，郑州大学网络空间安全学院，讲师，研究方向为医学图像分割、模式识别。
    本文中承担的工作：模型优化调整，实验数据整理。
    ZHAO Zhe is a lecturer at the School of Cyberspace Security, Zhengzhou University. Her research interests include medical image segmentation and pattern recognition.
    In this paper, she is mainly responsible for model optimization and adjustment, as well as experimental data organization.
    E-mail: sevenzz@zzu.edu.cn
  • 基金资助:
    国家自然科学基金(82441022);郑州市协同创新重大项目(20XTZX05015)

PMDeUNet: a Lightweight Network Model for the Efficient Segmentation of Abdominal Medical Image

LIN Yusong1,2(),LIANG Zhenyu1,ZHAO Zhe()   

  1. 1 School of Cyber Science and Engineering, Zhengzhou University, Zhengzhou, Henan 450002, China
    2 Collaborative Innovation Center for Internet Medical and Healthcare Service of Henan, Zhengzhou University, Zhengzhou, Henan 450052, China
  • Received:2025-11-04 Online:2026-08-20 Published:2026-08-21

RichHTML

6

PDF

23

摘要：

【目的】 设计一个可以部署在临床医疗设备上的高效轻量级腹部医学图像分割网络模型。【方法】 灵活结合CNN与Mamba架构的优势，构建出一个U形轻量级网络模型PMDeUNet。我们构造一个并行V-Mamba结构，将其作为模型的主干，以降低模型的复杂度并提升推理速度。同时设计一个并行多尺度可变形卷积模块并将其集成到模型中，针对性地提升腹部图像中形态复杂的器官的分割精度。【结果】 在公开数据集上的实验结果表明，我们的模型在保持理想的平均分割精度的同时，具有更少的参数量与更快的推理效率，并且对复杂器官的分割能力有进一步提升。【结论】 实验结果证明我们提出的模型结构在提高腹部器官分割精度与降低模型复杂度方面的有效性。

关键词: 腹部医学图像分割, 深度学习, Mamba 架构, 轻量级, 可变形卷积

Abstract:

[Objective] This study aims to design an efficient and lightweight abdominal medical image segmentation network model that can be deployed on clinical medical equipment. [Methods] The advantages of CNN and Mamba architectures are flexibly combined to construct a U-shaped lightweight network model PMDeUNet. A parallel V-Mamba structure is designed as the backbone of the model to reduce the model’s complexity and improve inference speed. Meanwhile, a parallel multi-scale deformable convolution module is integrated into the model to specifically enhance the segmentation accuracy of organs with complex morphologies in abdominal images. [Results] Experimental results on public datasets show that the proposed model maintains ideal average segmentation accuracy while having fewer parameters and higher inference efficiency, and the segmentation performance for complex organs is further improved. [Conclusions] The experimental results prove the effectiveness of the proposed model structure in improving the segmentation accuracy of abdominal organs and reducing model complexity.

Key words: abdominal medical image segmentation, deep learning, mamba architecture, lightweight, deformable convolution