基于W-Net的肝静脉和肝门静脉全自动分割

doi:10.3969/j.issn.0258-8021.2019.05.001

Abstract
Figure/Table
References
Related Citation (15)

Download: PDF (7569 KB) HTML (1 KB)
Export: BibTeX | EndNote (RIS)

Abstract Automatic segmentation of hepatic vein and portal vein based on preoperative CT images has important clinical value for liver segment. However, in the venous CT image of the liver, little difference is in the density of the hepatic vein and portal vein, and the vascular structure is also complicated, so there has been a challenge with automatic extraction of the hepatic vein and portal vein in 3D. To solve this problem, this paper proposed a network architecture named W-Net based on convolutional neural network (CNN). The network architecture made a full use of the difference in the 3D structure for the hepatic vein and portal vein. We also set a loss function for the extraction of all vessels and portal vein. Then the model would optimize the weighted sum of the two loss functions to automatically learn the features of all vessels and portal vein to achieve the best extraction result. Hepatic vein was obtained by subtracting the two result. In this paper, we used 10 sets of venous abdominal CT images from the public data set 3Dircadb01 for network model construction, and 10 groups for testing. We mainly used the Dice coefficient as the evaluation standard. Finally, in the liver area, the Dice coefficient and accuracy of blood vessels reached 0.715 and 0.970, the Dicecoefficient and accuracy of hepatic vein reached 0.597 and 0.984, and the Dice and accuracy of portal vein reached 0.608 and 0.970. We also tested in 10 sets of clinical data, and the method could effectively separate the hepatic vein and portal vein. The experimental results showed that the proposed method possessed better feature extraction ability and generalization ability, and had good performance in public data and clinical data.

Key words： hepatic vein portal vein segmentation convolutional neural networks

Received: 14 March 2019

PACS:

R318

	Service

	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Sun Jinfeng
	Ding Hui
	Wang Guangzhi

Cite this article:

Sun Jinfeng,Ding Hui,Wang Guangzhi. Automatic Segmentation of Hepatic Vein and Portal Vein Based on W-Net[J]. Chinese Journal of Biomedical Engineering, 2019, 38(5): 513-521.

URL:

http://cjbme.csbme.org/EN/10.3969/j.issn.0258-8021.2019.05.001 OR http://cjbme.csbme.org/EN/Y2019/V38/I5/513

[1] Couinaud C. Liver Anatomy: Portal (and Suprahepatic) or Biliary Segmentation[J]. Digestive Surgery, 1999, 16(6): 459-467.
[2] Selle D, Preim B, Schenk A, et al. Analysis of vasculature for liver surgical planning[J]. IEEE Transactions on Medical Imaging, 2002, 21(11): 1344-1357.
[3] Esneault S, Lafon C, Dillenseger JL. Liver vessels segmentation using a hybrid geometrical moments/graph cuts method[J]. IEEE Transactions on Biomedical Engineering, 2010, 57(2): 276-283.
[4] Chi Yanling, Liu Jimin, Venkatesh SK, et al. Segmentation of liver vasculature from contrast enhanced CT images using context-based voting[J]. IEEE Transactions on Biomedical Engineering, 2011, 58(8): 2144-2153.
[5] Krol A, Gimi B, Kitrungrotsakul T, et al. Robust hepatic vessel segmentation using multi deep convolution network[C] //Medical Imaging 2017: Biomedical Applications in Molecular, Structural, and Functional Imaging. Orlando: SPIE Medical Imaging, 2017:1013711.
[6] Ibragimov B, Toesca D, Chang D, et al. Combining deep learning with anatomical analysis for segmentation of the portal vein for liver SBRT planning[J]. Physics in Medicine & Biology, 2017, 62(23): 8943-8958.
[7] Huang Qing, Sun Jinfeng, Ding Hui, et al. Robust liver vessel extraction using 3D U-Net with variant Dice loss function[J]. Computers in Biology and Medicine, 2018, 101: 153-162.
[8] Huang Qing, Ding Hui, Wang Xiaodong, et al. Fully automatic liver segmentation in CT images using modified graph cuts and feature detection[J]. Computers in Biology and Medicine, 2018, 95: 198-208.
[9] Ronneberger O, Fischer P, Brox T. U-Net: Convolutional networks for biomedical image segmentation[C] // Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015. Munich: Springer, Cham, 2015: 234-241.
[10] içek Ö, Abdulkadir A, Lienkamp SS, et al. 3D U-Net: learning dense volumetric segmentation from sparse annotation[C] // Medical Image Computing and Computer-Assisted Intervention–MICCAI 2016. Athens: Springer, Cham, 2016: 424-432.
[11] Huang Gao, Liu Zhuang, Van DML, et al. Densely connected convolutional networks[C] // Proceedings of the IEEE conference on computer vision and pattern recognition. Honolulu: IEEE, 2017: 4700-4708.
[12] Milletari F, Navab N, Ahmadi SA. V-net: Fully convolutional neural networks for volumetric medical image segmentation[C] // 2016 Fourth International Conference on 3D Vision (3DV). Stanford: IEEE, 2016: 565-571.
[13] Bruyninckx P, Loeckx D, Vandermeulen D, et al. Segmentation of liver portal veins by global optimization[C] // Medical Imaging 2010: Computer-Aided Diagnosis. San Diego: SPIE Medical Imaging, 2010: 7624.
[14] Zhang Rui, Zhou Zhuhang, Wu Weiwei, et al. An improved fuzzy connectedness method for automatic three-dimensional liver vessel segmentation in CT images[J]. Journal of Healthcare Engineering, 2018, 2018: 1-18.