[1]李佳柠,王瑜,肖洪兵,等.基于改进U-Net网络的肾脏肿瘤全自动分割[J].中国医学物理学杂志,2023,40(10):1241-1245.[doi:DOI:10.3969/j.issn.1005-202X.2023.10.009]
　LI Jianing,WANG Yu,XIAO Hongbing,et al.Fully automated kidney tumor segmentation based on improved U-Net[J].Chinese Journal of Medical Physics,2023,40(10):1241-1245.[doi:DOI:10.3969/j.issn.1005-202X.2023.10.009]
点击复制

基于改进U-Net网络的肾脏肿瘤全自动分割()
分享到：

《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]

卷:
40卷
期数:
2023年第10期
页码:
1241-1245
栏目:
医学影像物理
出版日期:
2023-10-27

文章信息/Info

Title:
Fully automated kidney tumor segmentation based on improved U-Net
文章编号:
1005-202X（2023）10-1241-05
作者:
李佳柠; 王瑜; 肖洪兵; 闫善武; 孙梅
北京工商大学人工智能学院， 北京 100048
Author(s):
LI Jianing;  WANG Yu;  XIAO Hongbing;  YAN Shanwu;  SUN Mei
School of Artificial Intelligence, Beijing Technology and Business University, Beijing 100048, China
关键词:
U-Net网络; CT影像; 肾脏肿瘤; 多尺度卷积; 残差模块
Keywords:
Keywords: U-Net CT image kidney tumor multi-scale convolution residual module
分类号:
1005-202X（2023）10-1241-05
DOI:
DOI:10.3969/j.issn.1005-202X.2023.10.009
文献标志码:
A
摘要:
针对肾脏肿瘤大小、位置不确定以及传统U-Net网络全自动分割肾脏肿瘤时易出现过拟合等难题，提出一种改进的多尺度卷积和残差U-Net（MSR U-Net）的肾脏肿瘤全自动分割方法。一方面，在残差模块中加入跳跃连接使网络收敛得更快，缓解过拟合现象；另一方面，在多尺度卷积模块中采用3种不同尺寸的卷积核，增加网络的感受野，解决网络提取的肿瘤特征尺寸单一问题。使用KITS19数据库中90例患者的CT切片进行相关验证性实验，MSR U-Net方法获得了肾脏的Dice系数为0.976和肿瘤的Dice系数为0.836，表明MSR U-Net在全自动肾脏肿瘤分割任务中的可行性和有效性。
Abstract:
Abstract: Aiming at some intractable problems such as the uncertain size and location of kidney tumors, and overfitting etc. when using the traditional U-Net network to segment kidney tumor full-automatically, a fully automated segmentation method for kidney tumors based on multi-scale and residual U-Net (MSR U-Net) is proposed. Skip connections is added in the residual module to make the network converge faster, further avoiding the overfitting. Moreover, 3 different sizes of convolution kernels are used in the multi-scale convolution module to increase the receptive field, solving the problem of single feature size. The CT slices of 90 patients in the KITS19 database are used to carry out relevant verification experiments. MSR U-Net method obtains a Dice coefficient of 0.976 for kidney and 0.836 for tumor, indicating that the proposed algorithm is feasible and effective in the fully automated kidney tumor segmentation task.

相似文献/References:

[1]裴潇倜,吕琳,黄鹏杰,等.基于U-Net的T细胞斑点检测方法研究[J].中国医学物理学杂志,2021,38(4):518.[doi:DOI:10.3969/j.issn.1005-202X.2021.04.022]
　PEI Xiaoti,L?Lin,HUANG Pengjie,et al.T-cell spot test based on U-Net[J].Chinese Journal of Medical Physics,2021,38(10):518.[doi:DOI:10.3969/j.issn.1005-202X.2021.04.022]
[2]孙旭阳,石更强.基于CT影像的人体踝关节三维重建[J].中国医学物理学杂志,2022,39(7):850.[doi:DOI:10.3969/j.issn.1005-202X.2022.07.010]
　SUN Xuyang,SHI Gengqiang.Three-dimensional reconstruction of the ankle joint based on CT image[J].Chinese Journal of Medical Physics,2022,39(10):850.[doi:DOI:10.3969/j.issn.1005-202X.2022.07.010]
[3]王萍,徐凯成,张一弛,等.基于反注意力机制U-Net网络的胃部肿瘤分割[J].中国医学物理学杂志,2022,39(9):1133.[doi:DOI:10.3969/j.issn.1005-202X.2022.09.013]
　WANG Ping,XU Kaicheng,ZHANG Yichi,et al.Gastric tumor segmentation by U-Net based on reverse attention mechanism[J].Chinese Journal of Medical Physics,2022,39(10):1133.[doi:DOI:10.3969/j.issn.1005-202X.2022.09.013]
[4]韩伟,李志雄,郝英剑,等.基于CT影像的人体心脏三维重构[J].中国医学物理学杂志,2023,40(2):170.[doi:DOI:10.3969/j.issn.1005-202X.2023.02.007]
　HAN Wei,LI Zhixiong,HAO Yingjian,et al.Three-dimensional reconstruction of the heart from CT images[J].Chinese Journal of Medical Physics,2023,40(10):170.[doi:DOI:10.3969/j.issn.1005-202X.2023.02.007]

备注/Memo

备注/Memo:
【收稿日期】2023-03-14 【基金项目】北京市自然科学基金-北京市教育委员会科技计划重点项目（KZ202110011015） 【作者简介】李佳柠，硕士，研究方向：医学图像处理与模式识别，E-mail: 990670042@qq.com 【通信作者】王瑜，教授，研究方向：图像处理与模式识别，E-mail: wangyu@btbu.edu.cn
更新日期/Last Update: 2023-10-27