A lightweight MRI image segmentation method based on MobileNetV3 and SwinTransformer

By combining a lightweight CNN and an improved TransUNet model with Transformer, the accuracy and efficiency issues of MRI image segmentation on small datasets are solved, achieving efficient and accurate tumor region and boundary segmentation, which is suitable for automated clinical diagnosis.

CN122416005APending Publication Date: 2026-07-17GUILIN UNIV OF ELECTRONIC TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
GUILIN UNIV OF ELECTRONIC TECH
Filing Date
2025-08-25
Publication Date
2026-07-17

Smart Images

  • Figure CN122416005A_ABST
    Figure CN122416005A_ABST
Patent Text Reader

Abstract

本发明公开了一种基于轻量化改进TransUNet的MRI图像分割方法，旨在解决现有TransUNet模型庞大、显存占用高、对小数据集适应性差及卷积与Transformer交互弱等问题。该方法以TransUNet为基础模型，通过多方面创新优化实现高效分割：采用MobileNetV3 Small作为轻量化CNN Backbone，减少参数量5~10倍以适配小数据集；引入Swin Transformer Tiny作为轻量Transformer Encoder，利用分层窗口注意力降低计算复杂度，增强局部与全局信息捕获能力；添加Neck ASPPLight模块以增强多尺度上下文感受野，补充全局信息；设计LightCUPHead作为Decoder，通过深度可分离卷积和1×1卷积通道映射降低参数量并保证通道对齐。改进后的模型轻量快速，在小数据集上表现出更优的收敛性和分割精度，为医学MRI图像分割提供了高效可靠的技术支持，适用于医学诊断与治疗辅助等领域。
Need to check novelty before this filing date? Find Prior Art