ScholarMate
English简体中文繁體中文
客服热线:400-1616-289
登录注册
会议论文

LTUNet: A Lightweight Transformer-Based UNet with Multi-scale Mechanism for Skin Lesion Segmentation

Guo, Huike; Zhang, Han; Li, Minghe; Quan, Xiongwen*南开大学
3rd CAAI International Conference on Artificial Intelligence (CICAI)2024DOI: 10.1007/978-981-99-9119-8_14
1阅读
引用收藏
分享

摘要

Medical image segmentation separates target structures or tissues within medical images to promote precise diagnoses. Automated image segmentation algorithms can help dermatologists to diagnose skin cancer by identifying skin lesions. Many popular image segmentation algorithms combine UNet and Transformer, but cannot fully utilize the global information between different scales and also have a huge number of parameters. To this end, this paper proposes a lightweight Transformer-based UNet (LTUNet) method for medical image segmentation, which designs an effective approach to extract and fully use multi-scale features. Firstly, the multi-scale feature maps of images are extracted by the inverted residual blocks of lightweight UNet encoder. Then, the feature maps are concatenated as the input of the Transformer's encoder blocks to compute intra- and inter-scale attention scores, and the scores are used to enhance the feature map of each scale. Finally, we fuse the upsampled results of all scales on UNet to improve the performance of segmentation. Our method achieves 0.9432, 0.8948, 0.9348 for mDice, mIoU and mACC on the ISIC2016 dataset, and 0.9058, 0.8138, 0.8968 on the ISIC2018 dataset respectively, which outperforms state-of-the-art methods. Besides, our network has a smaller number of parameters and converges faster.

学科与关键词

关键词
Medical image segmentationUNetTransformerMulti-scaleSkin lesion
相关
加载中…
  • SUTrans-NET: a hybrid transformer approach to skin lesion segmentation
    Li, Yaqin; Tian, Tonghe; Hu, Jing; Yuan, Cao*PeerJ Computer Science, 2024, 10: e1935.
  • Semi-Supervised Skin Lesion Segmentation With Coupling CNN and Transformer Features
    Alahmadi, Mohammad D. D.*; Alghamdi, WajdiIEEE Access, 2022, 10: 122560-122569.被引 9
  • SkinFormer: Learning Statistical Texture Representation With Transformer for Skin Lesion Segmentation
    Xu, Rongtao; Wang, Changwei; Zhang, Jiguang; Xu, Shibiao*; Meng, Weiliang*; Zhang, XiaopengIEEE Journal of Biomedical and Health Informatics, 2024, 28(10): 6008-6018.被引 2
查看全部 →

会议信息

会议名称
3rd CAAI International Conference on Artificial Intelligence (CICAI)
会议日期
2023-07-22 至 2023-07-23
论文集名
ARTIFICIAL INTELLIGENCE, CICAI 2023, PT II
发表日期
2024
起止页码
147-158
DOI
10.1007/978-981-99-9119-8_14
资助标注
Major Program of the National Social Science Foundation of China [21ZD102]
收录
ISTP

作者

  • G

    Guo, Huike (第一作者)

    nankai univ

  • Z

    Zhang, Han

    nankai univ

查看全部 4 位作者收起作者
  • L

    Li, Minghe

    nankai univ

  • Q

    Quan, Xiongwen (通讯)

    nankai univ

登录解锁更多功能

  • 收藏这篇,随时在“我的 · 收藏”找回
  • 认领你的成果,生成推广素材
  • 关注作者获取最新成果
  • 加入项目组、协同写作
注册 / 登录

产品服务

  • 科研之友
  • 创新城
  • 科创云

服务支持

  • 帮助中心
  • 隐私政策
  • 服务条款

联系方式

在线客服:【】
客服热线:400-1616-289
电子邮箱:support@scholarmate.com

关注或下载科研之友

微信二维码
微信公众号
客户端下载二维码
下载客户端
科研成果科研人员科研机构科研动态爱瑞思软件

©2026 深圳市科研之友网络服务有限公司

公安备案图标粤公网安备 44030502000213
粤ICP备 16046710 号粤B2-20110417