Please wait a minute...
浙江大学学报（理学版）
浙江大学学报（理学版）  2020, Vol. 47 Issue (1): 20-26    DOI: 10.3785/j.issn.1008-9497.2020.01.003
人工智能与可视计算     
DenseNet-centercrop: 一个用于肺结节分类的卷积网络
刘一璟1,3, 张旭斌1, 张建伟1, 周哲磊1,3, 冯元力1, 陈为1,2
1.浙江大学 计算机科学与技术学院 CAD&CG国家重点实验室，浙江杭州 310058
2.浙江大学 附属第一医院，浙江杭州 310006
3.浙江大学 数学科学学院，浙江杭州 310027
DenseNet-centercrop: A novel convolutional network for lung nodule classification
LIU Yijing1,3, ZHANG Xubin1, ZHANG Jianwei1, ZHOU Zhelei1,3, FENG Yuanli1, CHEN Wei1,2
1.State Key Laboratory of CAD&CG, College of Computer Science and Technology, Zhejiang University, Hangzhou 310058, China
2.The First Affiliated Hospital, Zhejiang University, Hangzhou 310006, China
3.School of Mathematical Sciences, Zhejiang University, Hangzhou 310027, China
 全文: PDF(819 KB)   HTML  
摘要： 为解决由肺部CT图像对肺结节进行良恶性分类的问题， 提出了一个新颖的端到端深度学习网络DenseNet-centercrop。通过在原有的DenseNet结构中的稠密块间增加新的分支，引入了中心剪裁操作。该网络结构具有2个优势: （1）不仅最大程度保留了DenseNet的结构，而且将其稠密连接机制扩展到了稠密块水平，大大丰富了肺结节的多尺度特征。（2）参数量较少，是一种轻量化的网络结构。将基于该网络的肺结节良恶性分类方法在LIDC-IDRI数据集上进行评估， 实验结果表明， DenseNet-centercrop极大地提高了DenseNet的性能， 较现有的其他肺结节良恶性分类方法具有更高的AUC分值和分类精度。
关键词： 肺结节分类;  电子计算机断层扫描图像;  稠密连接卷积网络    
Abstract: To solve the problem of benign or malignant diagnosis of pulmonary nodule with original thoracic computed tomography (CT) images, this paper presents a novel end-to-end deep learning architecture named DenseNet-centercrop. DenseNet-centercrop has two compelling advantages: (1) DenseNet-centercrop preserves the structure of DenseNet at utmost, further reinforces the densely connected mechanism to a level of dense blocks and enriches multi-scale features of lung nodules. (2) It is a lightweight structure with small scale of parameters. We evaluate DenseNet-centercrop on LIDC-IDRI benchmark. Experimental results show that DenseNet-centercrop not only largely boosts the performance of DenseNet, it also has higher accuracy and AUC score on the task of lung nodule classification in comparison with state-of-the-art approaches.
Key words: lung nodule classification    computed tomography (CT) imaging    densely connected convolutional networks
收稿日期: 2019-08-29 出版日期: 2020-01-25
CLC:  TP391.41  
基金资助: 国家自然科学基金资助项目（61772456）；浙江大学教育基金项目（K18-51120-004， K17-51120-017）.
通讯作者: ORCID:http://orcid.org/0000-0001-7845-1379，E-mail: chenwei@cad.zju.edu.cn.   
作者简介: 刘一璟(1998—)，ORCID:http://orcid.org/0000-0001-8420-2213，男，博士研究生，主要从事深度学习、医学图像处理等研究.
服务　　
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章　　
刘一璟
张旭斌
张建伟
周哲磊
冯元力
陈为

引用本文:

刘一璟, 张旭斌, 张建伟, 周哲磊, 冯元力, 陈为. DenseNet-centercrop: 一个用于肺结节分类的卷积网络[J]. 浙江大学学报（理学版）, 2020, 47(1): 20-26.

LIU Yijing, ZHANG Xubin, ZHANG Jianwei, ZHOU Zhelei, FENG Yuanli, CHEN Wei. DenseNet-centercrop: A novel convolutional network for lung nodule classification. Journal of Zhejiang University (Science Edition), 2020, 47(1): 20-26.

链接本文:

https://www.zjujournals.com/sci/CN/10.3785/j.issn.1008-9497.2020.01.003    或    https://www.zjujournals.com/sci/CN/Y2020/V47/I1/20

1 SIEGEL R L , MILLER K D . Cancer statistics[J].CA: A Cancer Journal for Clinicians，2018，68 (1) : 7–30.
2 LAMBIN P , RIOS-VELAZQUEZ E . Radiomics: Extracting more information from medical images using advanced feature analysis[J]. European Journal of Cancer, 2012,48 (4) : 441–446.DOI:10.1016/j.ejca.2011.11.036
3 ARMATO S G , ALTMAN M B . Automated lung nodule classification following automated nodule detection on CT: A serial approach[J]. Medical Physics ,2003,30 (6) : 1188–1197.DOI:10.1118/1.1573210
4 VALENTE I R S , CORTEZ P C . Automatic 3d pulmonary nodule detection in CT images: A survey[J]. Computer Methods and Programs in Biomedicine, 2016,124 : 91–107.DOI:10.1016/j.cmpb.2015.10.006
5 AERTS H J W L , VELAZQUEZ E R . Decoding tumour phenotype by noninvasive imaging using a quantitative radiomics approach[J]. Nature Communications, 2014,5 (1): 4006.DOI:10.1038/ncomms5006
6 BALAGURUNATHAN Y , GU Y . Reproducibility and prognosis of quantitative features extracted from CT images[J]. Translational Oncology, 2014,7 (1): 72–87. DOI:10.1593/tlo.13844
7 KUMAR D , WONG A , DAVID A . Lung nodule classification using deep features in CT images[C]// 12th Conference on Computer and Robot Vision.Halifax，NS：IEEE, 2015, 133-138. DOI:10.1109/CRV.2015.25
8 OROZCO H M , VILLEGAS O O V . Automated system for lung nodules classification based on wavelet feature descriptor and support vector machine[J]. BioMedical Engineering Online, 2015, 14 (1): 9.DOI:10.1186/s12938-015-0003-y
9 WAY T W, HADJIISKI L M . Computer-aided diagnosis of pulmonary nodules on CT scans: Segmentation and classification using 3d active contours[J]. Medical Physics, 2006, 33 (7Part1): 2323–2337. DOI:10.1118/1.2207129
10 SONG J , YANG C . Lung lesion extraction using a toboggan based growing automatic segmentation approach[J]. IEEE Transactions on Medical Imaging, 2016, 35 (1): 337–353.DOI:10.1109/TMI.2015.2474119
11 EL-BAZ A , NITZKEN M . 3D shape analysis for early diagnosis of malignant lung nodules[C]//Biennial International Conference on Information Processing in Medical Imaging. Berlin /Heidelberg： Springer, 2011, 772–783. DOI:10.1007/978-3-642-22092-0_63
12 HAN F , WANG H . Texture feature analysis for computer-aided diagnosis on pulmonary nodules[J]. Journal of Digital Imaging, 2015, 28 (1): 99–115.DOI:10.1007/s10278-014-9718-8
13 CIOMPI F , DE HOOP B . Automatic classification of pulmonary peri-fissural nodules in computed tomography using an ensemble of 2d views and a convolutional neural network out-of-the-box[J]. Medical Image Analysis, 2015, 26 (1): 195–202.
14 DONG C , YIN Y . Detecting malignant patients via modified boosted tree[J]. Science China Information Sciences, 2010, 53 (7): 1369–1378.DOI:10.1007/s11432-010-3107-9
15 ALILOU M , OROOJI M . Intra-perinodular textural transition (ipris): A 3d descriptor for nodule diagnosis on lung CT[C]// International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2017, 647–655. DOI:10.1007/978-3-319-66179-7_74
16 FARAG A A , ALI A . Feature fusion for lung nodule classification[J]. International Journal of Computer Assisted Radiology and Surgery, 2017, 12 (10) : 1809–1818. DOI:10.1007/s11548-017-1626-1
17 CIOMPI F , CHUNG K . Towards automatic pulmonary nodule management in lung cancer screening with deep learning[J]. Scientific Reports, 2017, 7: 46479. DOI:10.1038/srep46479
18 ZHAO X , LIU L . Agile convolutional neural network for pulmonary nodule classification using CT images[J]. International Journal of Computer Assisted Radiology and Surgery, 2018, 13 (4): 585–595.DOI:10.1007/s11548-017-1696-0
19 QIANG Y , GE L . Pulmonary nodule diagnosis using dual-modal supervised autoencoder based on extreme learning machine[J]. Expert Systems, 2017, 34 (6): e12224. DOI:10.1111/exsy.12224
20 HUANG G , LIU Z . Densely connected convolutional networks[C]// Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu,HI: IEEE, 2017, 4700-4708. DOI:10.1109/cvpr.2017.243
21 SHEN W , ZHOU M . Multi-scale convolutional neural networks for lung nodule classification[C] // International Conference on Information Processing in Medical Imaging. Cham: Springer, 2015, 9123: 588–599. DOI:10.1007/978-3-319-19992-4_46
22 KINGMA D P , BA J . Adam: A Method for Stochastic Optimization[C]. arXiv preprint arXiv:1412.6980, 2014.
23 LIN T Y , GOYAL P . Focal loss for dense object detection[C]// Proceedings of the IEEE International Conference on Computer Vision (ICCV). Venice: IEEE, 2017, 2999–3007. DOI:10.1109/TPAMI.2018.2858826
24 HUSSEIN S , GILLIES R . TumorNet: Lung nodule characterization using multi-view convolutional neural network with gaussian process[C] //2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017. Melbourne,VIC: IEEE, 2017, 1007–1010. DOI:10.1109/isbi.2017.7950686
25 BUTY M , XU Z . Characterization of lung nodule malignancy using hybrid shape and appearance features[C]// International Conference on Medical Image Computing and Computer-Assisted Intervention.Cham: Springer , 2016, 662–670. DOI:10.1007/978-3-319-46720-7_77
26 SHEN W , ZHOU M . Multi-crop convolutional neural networks for lung nodule malignancy suspiciousness classification[J]. Pattern Recognition, 2017, 61: 663–673.DOI:10.1016/j.patcog.2016.05.029
[1] 罗月童, 韩承村, 杜华, 严伊蔓. 基于拉伸特征的B-Rep→CSG转换算法及其应用[J]. 浙江大学学报（理学版）, 2021, 48(2): 151-158.
[2] 陈佳舟, 王宇航, MohammedAmal Ahmed Hasan, 黄可妤, 卢周扬, 彭群生. 基于图像的二维剪纸自动生成方法[J]. 浙江大学学报（理学版）, 2020, 47(3): 274-283.
[3] 卢家品, 罗月童, 黄兆嵩, 张延孔, 陈为. 基于排名学习和多源信息的地图匹配方法[J]. 浙江大学学报（理学版）, 2020, 47(1): 27-35.
[4] 李丽, 高若婉, 梅树立, 赵海英. 基于Shannon-Cosine小波精细积分法的壁画降噪修复方法[J]. 浙江大学学报（理学版）, 2019, 46(3): 279-287.
[5] 傅红普, 邹北骥. 一种方向梯度直方图的降维方法[J]. 浙江大学学报（理学版）, 2017, 44(2): 134-138.
[6] 张莉, 赵林, 檀结庆. 带互异权值的渐进迭代逼近算法及其应用[J]. 浙江大学学报（理学版）, 2017, 44(1): 22-27.
[7] 桂彦, 王培玉, 李峰, 刘杨. 基于GPU加速的几何纹理合成方法[J]. 浙江大学学报（理学版）, 2016, 43(6): 638-646.
[8] 刘玉杰, 庞芸萍, 李宗民, 李华. 融合抽象层级变换和卷积神经网络的手绘图像检索方法[J]. 浙江大学学报（理学版）, 2016, 43(6): 657-663.