摘 要：针对胸腔CT 影像信息复杂度高、肺器官体积大造成的3D 肺实质结构无法快速准确分割的问题，提出了一种基于改进U-net 网络的2.5D 肺实质分割方法。将3 个轴向（冠状面、矢状面、横截面）的胸腔CT 影像分别输入改进U-net 网络模型进行特征学习，而后对3 个轴向的学习结果进行融合，实现3D 肺实质分割。使用该方法在中南民族大学认知科学实验室中完成了一系列肺实质分割实验，实验结果表明该方法可以有效地完成3D 肺实质分割。
关键词：肺实质分割；改进U-net 网络；2.5D
中图分类号：TP391.41；R734.2 文献标识码：A 文章编号：2096-4706（2020）09-0085-04
A 2.5D Lung Segmentation for CT Images Based on Improved U-net
WANG Nan，WANG Senmei，CAI Jing
（South-central University for Nationalities，Wuhan 430074，China）
Abstract：In order to solve the problem that 3D lung parenchyma structure cannot be quickly and accurately segmented due to the high complexity of chest CT image information and large lung organ volume，a 2.5D lung-segmentation method based on improved U-net is proposed. Thoracic CT images in three axial directions （coronal，sagittal，transversal） were input into the improved U-net network model for feature learning，and then the three axial learning results were fused to achieve 3D lung-segmentation. A series of lung parenchyma segmentation experiments have been completed in the Cognitive Science Laboratory of South-central University for Nationalities. The experimental results show that this method can effectively complete 3D lung lung-segmentation.
Keywords：lung-segmentation；improved U-net；2.5D
参考文献：
[1] HU S，HOFFMAN EA，REINHARDT JM. Automatic lungsegmentation for accurate quantitation of volumetric X-ray CT images [J].IEEE transactions on medical imaging，2001，20（6）：490-498.
[2] EVAN S，JONATHAN L，TREVOR D. Fully ConvolutionalNetworks for Semantic Segmentation [J]. IEEE transactions on patternanalysis and machine intelligence，2017，39（4）：640-651.
[3] RONNEBERGER O，FISCHER P，BROX T. U-Net：Convolutional Networks for Biomedical Image Segmentation [C]//18thInternational Conference，October 5-9，2015，Proceedings，Part III.Munich，Germany：Medical Image Computing and Computer-Assisted Intervention-MICCAI，2015：234-241.
[4] HE K M，ZHANG X Y，REN S Q，et al.Identity Mappingsin Deep Residual Networks [C]//14th European Conference，October11-14，2016，Proceedings，Part IV.Netherlands：ECCV 2016：Computer Vision-ECCV，2016：630-645.
[5] SOLIMAN A，SHAFFIE A，GHAZAL M，et al. A Novel CNN Segmentation Framework Based on Using New Shape and Appearance Features [C]//IEEE：2018 25th IEEE International Conference on Image Processing （ICIP），2018：3488-3492.
[6] GU Z W，CHENG J，FU H Z，et al. CE-Net：Context Encoder Network for 2D Medical Image Segmentation [J]. IEEE Transactions on Medical Imaging，2019：1.
[7] HE K M，ZHANG X Y，REN S Q，et al. Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition [J].IEEE transactions on pattern analysis and machine intelligence，2015，37（9）：1904-1916.
[8] 顾一凡. 基于分组膨胀卷积和级联网络的语义分割算法研究 [D]. 哈尔滨：哈尔滨工业大学，2018.
[9] 曹玉磊.DICOM 标准研究与图像处理工具的实现 [D]. 西安：西安电子科技大学，2007.
[10] 赵向明. 基于深度学习的医学图像分割算法研究 [D]. 武汉：华中科技大学，2019.
作者简介：王楠（1994-），女，蒙古族，河南南阳人，硕士研究生，研究方向：医学图像处理。