《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (3): 818-824.DOI: 10.11772/j.issn.1001-9081.2021040948

• 2021年中国计算机学会人工智能会议(CCFAI 2021) • 上一篇    

基于注意力机制的多尺度残差UNet实现乳腺癌灶分割

罗圣钦1, 陈金怡1, 李洪均1,2()   

  1. 1.南通大学 信息科学技术学院,江苏 南通 226019
    2.计算机软件新技术国家重点实验室(南京大学),南京 210023
  • 收稿日期:2021-06-04 修回日期:2021-06-22 接受日期:2021-06-29 发布日期:2021-11-09 出版日期:2022-03-10
  • 通讯作者: 李洪均
  • 作者简介:罗圣钦(1997—),男,江苏盐城人,硕士研究生,CCF会员,主要研究方向:深度学习、医学图像处理
    陈金怡(1996—),男,江苏苏州人,硕士研究生,CCF会员,主要研究方向:深度学习、计算机视觉;
  • 基金资助:
    国家自然科学基金资助项目(61976120);江苏省产学研合作项目(BY2021349);江苏省研究生科研与实践创新计划项目(KYCX21_3084);南通市科技计划资助项目(JC2021131);南京大学计算机软件新技术国家重点实验室基金资助项目(KFKT2019B015)

Multiscale residual UNet based on attention mechanism to realize breast cancer lesion segmentation

Shengqin LUO1, Jinyi CHEN1, Hongjun LI1,2()   

  1. 1.School of Information Science and Technology,Nantong University,Nantong Jiangsu 226019,China
    2.State Key Laboratory for Novel Software Technology (Nanjing University),Nanjing Jiangsu 210023,China
  • Received:2021-06-04 Revised:2021-06-22 Accepted:2021-06-29 Online:2021-11-09 Published:2022-03-10
  • Contact: Hongjun LI
  • About author:LUO Shengqin, born in 1997, M. S. candidate. His research interests include deep learning, medical image processing.
    CHEN Jinyi, born in 1996, M. S. candidate. His research interests include deep learning, computer vision.
  • Supported by:
    National Natural Science Foundation of China(61976120);Jiangsu Industry University Research Cooperation Project(BY2021349);Postgraduate Research & Practice Innovation Program of Jiangsu Province(KYCX21_3084);Science and Technology Program of Nantong(JC2021131);State Key Laboratory for Novel Software Technology at Nanjing University(KFKT2019B015)

摘要:

针对乳腺癌灶在磁共振成像(MRI)中呈现大小形状不一、边界模糊等特点,为避免误分割并提高分割精度,提出一种基于注意力机制的多尺度残差UNet分割算法。首先,利用多尺度残差单元替换UNet在下采样过程中的相邻两个卷积块以加强对形态大小差异的关注;接着,在上采样阶段使用跨层的注意力引导网络对重点区域的关注,避免造成对健康组织的误分割;最后,引入空洞空间金字塔池化作为分割网络的桥接模块以强化对病灶的表征能力。与UNet相比,所提算法在Dice系数、交并比(IoU)、特异度(SP)、准确度(ACC)等指标上分别提升了2.26、2.11、4.16、0.05个百分点。实验结果表明,所提算法能够提高癌灶分割精度,有效降低影像诊断的假阳性率。

关键词: 乳腺癌灶分割, 多尺度残差, 注意力机制, 桥接模块, 假阳性率

Abstract:

Concerning the characteristics of breast cancer in Magnetic Resonance Imaging (MRI), such as different shapes and sizes, and fuzzy boundaries, an algorithm based on multiscale residual U Network (UNet) with attention mechanism was proposed in order to avoid error segmentation and improve segmentation accuracy. Firstly, the multiscale residual units were used to replace two adjacent convolution blocks in the down-sampling process of UNet, so that the network could pay more attention to the difference of shape and size. Then, in the up-sampling stage, layer-crossed attention was used to guide the network to focus on the key regions, avoiding the error segmentation of healthy tissues. Finally, in order to enhance the ability of representing the lesions, the atrous spatial pyramid pooling was introduced as a bridging module to the network. Compared with UNet, the proposed algorithm improved the Dice coefficient, Intersection over Union (IoU), SPecificity (SP) and ACCuracy (ACC) by 2.26, 2.11, 4.16 and 0.05 percentage points, respectively. The experimental results show that the algorithm can improve the segmentation accuracy of lesions and effectively reduce the false positive rate of imaging diagnosis.

Key words: breast cancer lesion segmentation, multiscale residual, attention mechanism, bridging module, false positive rate

中图分类号: