深度全卷积网络对MRI膀胱图像的分割

韩文忠; 康莉; 江静婉; 黄建军

doi:10.16798/j.issn.1003-0530.2019.03.016

深度全卷积网络对MRI膀胱图像的分割

深圳大学信息工程学院

基金项目: 广东省自然科学基金项目（2018A030310511）；深圳市科技计划项目（JCYJ20170301150224580）

详细信息

通讯作者:
康莉 E-mail: kangli@szu.edu.cn

中图分类号: TP391
计量
- 文章访问数: 166
- HTML全文浏览量: 16
- PDF下载量: 574
出版历程
- 收稿日期: 2018-12-30
- 修回日期: 2019-02-25
- 发布日期: 2019-03-24

Segmentation of MRI Bladder Images by Deep Fully Convolutional Network

College of Information Engineering, Shenzhen University

More Information

Corresponding author:
Kang Li E-mail: kangli@szu.edu.cn

摘要

摘要: 大多数膀胱癌患者的膀胱肿瘤组织和膀胱壁组织互相渗透，各自的大小、形状变化多样，位置不固定，且膀胱MRI（Magnetic Resonance Imaging）图像中存在复杂的噪声和伪影，这使得将肿瘤和膀胱壁两者精确分割出来为下一步治疗进行诊断和定量分析成为难题。文中提出一种以U-net作为基础网络框架的深度全卷积网络，用残差网络子模块代替普通的卷积层进行下采样，通过空洞卷积来提取特征图不同感受野的信息，从而对不同尺度的特征图进行并行分支下采样。针对数据集小的问题，提出对图像加入高斯噪声、调节亮度和各向异性扩散滤波三种方法来进行数据扩增。实验结果表明，文中提出的方法对肿瘤分割的DSC（Dice similarity coefficient）值达到了0.9058，对膀胱壁分割的DSC值达到了0.9038，能够达到很好的分割效果。
- 深度全卷积 /
- 残差网络 /
- 空洞卷积 /
- 下采样 /
- 各向异性扩散滤波
Abstract: Bladder tumor tissue and bladder wall tissue of most patients with bladder cancer penetrate each other, with variable size, shape, and position. And there are more complicated noises and artifacts in the image of bladder MRI. These maked the precise segmentation of both the tumor and the bladder wall becomes a problem for diagnosis and quantitative analysis of the next treatment. In this paper, a deep convolutional network with U-net as the basic network framework has been proposed. The residual network as a sub-module replaced the ordinary convolutional layer for downsampling, and the dilated convolution has been used to extract the information of different receptive fields in the feature map by parallel branch downsampling. Aiming at the problem with poor data sets, three methods of adding Gaussian noise, adjusting brightness and anisotropic diffusion filtering have been proposed for data augmentation. The experimental results shown that the method proposed in this paper got a DSC value of 0.9058 for bladder tumor, and the DSC value of the bladder wall segmentation reached 0.9038, which can achieve a good segmentation effect.
- deep fully convolution /
- residual network /
- dilated convolution /
- downsampling /
- anisotropic diffusion filtering

HTML全文

参考文献(17)

[1]	Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide: sources, methods and major patterns in GLOBOCAN 2012.[J]. International Journal of Cancer, 2015, 136(5):E359-E386.
[2]	Grasso M. Bladder Cancer: A Major Public Health Issue[J]. European Urology Supplements, 2008, 7(7):510-515.
[3]	Sajadi AS, Sabzpoushan SH. A New Seeded Region Growing Technique for Retinal Blood Vessels Extraction. Journal of medical signals and sensors. 2014 Jul;4(3):223.
[4]	CERNAZANUGLAVAN, and HOLBAN. "Segmentation of Bone Structure in X-ray Images using Convolutional Neural Network." Advances in Electrical & Computer Engineering 13.1(2013):87-94.
[5]	LeCun Y，Boser B，Denker J S，Henderson D，Howard R E，Hubbard W，Jackel L D．Backpropagation applied to handwritten zip code recognition．Neural Conputationon,1989，1(4)：541-551.
[6]	Long J，Shelhamer E，Darrll T．Fully convolutional networks for semantic segmentation．In： Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition．Boston，Massachusetts，USA：IEEE，2015．3431-3440.
[7]	Moeskops P, Wolterink J M, Velden B H M V D, et al. Deep Learning for Multi-task Medical Image Segmentation in Multiple Modalities[C]// International Conference on Medical Image Computing & Computer-assisted Intervention. 2016.
[8]	Ronneberger O，Fischer P，Brox T．U-net：convolutional networks for biomedical image segmentation．Medical Image Computing and Computer-Assisted Intervention．Munich，Germany：Springer，2015，9351：234-241.
[9]	Brosch T, Tang L, Yoo Y, et al. Deep 3D Convolutional Encoder Networks with Shortcuts for Multiscale Feature Integration Applied to Multiple Sclerosis Lesion Segmentation[J]. IEEE Transactions on Medical Imaging, 2016, 35(5):1229-1239.
[10]	肖丹, 刘洋, 张国鹏, et al. 基于厚度特征模糊C均值算法的膀胱肿瘤疑似区域三维分割方法研究[J]. 医疗卫生装备, 2015, 36(3):1-4.
[11]	Xiao Dan, Liu Yang, Zhang Guopeng, et al. Fuzzy C-means algorithm using wall thickness feature for 3D bladder tumor suspected region segmentation[J]. Chinese Medical Equipment Journal, 2015, 36(3):1-4.(in Chinese)
[12]	K. He, X. Zhang, S. Ren, and J. Sun. Deep residuallearning for image recognition. In CVPR, 2016.
[13]	Yu, Fisher, and Vladlen Koltun. Multi-scale context aggregation by dilated convolutions. arXiv preprint arXiv:1511.07122 (2015).
[14]	Tang J, Sun Q, Liu J, et al. An Adaptive Anisotropic Diffusion Filter for Noise Reduction in MR Images[C]// International Conference on Mechatronics & Automation. 2007.
[15]	Gao H, Zhuang L, Maaten L V D, et al. Densely Connected Convolutional Networks[C]// IEEE Conference on Computer Vision & Pattern Recognition. 2017.
[16]	Yu X, Efe M O, Kaynak O. A general backpropagation algorithm for feedforward neural networks learning[J]. IEEE Transactions on Neural Networks, 2002, 13(1):251.
[17]	Moeskops P, Viergever M A, Mendrik A M, et al. Automatic Segmentation of MR Brain Images With a Convolutional Neural Network[J]. IEEE Transactions on Medical Imaging, 2016, 35(5):1252-1261.

施引文献(5)

期刊类型引用(4)

1.	刘力宾，李翔，魏本征，张魁星. 基于Swin Transformer的多尺度边缘优化膀胱癌磁共振成像分割算法. 生物医学工程研究. 2023(01): 43-49 . 百度学术
2.	张娜，张永寿，李翔，丛金玉，李徐周，魏本征. 多尺度特征融合的轻量化膀胱癌MRI图像分割算法. 陕西师范大学学报(自然科学版). 2022(03): 89-95 . 百度学术
3.	许广，管军霖，甘才军，汪华登. 一种新的基于U-Net和ResNet的病理图像细胞核分割方法. 广西科学院学报. 2021(04): 372-379+400 . 百度学术
4.	刘敏，方志军，高永彬. 多阶段冠状动脉CTA三维分割算法. 信号处理. 2019(11): 1911-1918 . 本站查看