首页 > 论文 > 激光与光电子学进展 > 55卷 > 5期(pp:51006--1)

二维和三维卷积神经网络相结合的CT图像肺结节检测方法

Detection of Pulmonary Nodules CT Images Combined with Two-Dimensional and Three-Dimensional Convolution Neural Networks

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

针对现有方法在大量肺部数据中存在的检测肺结节效率不高及大量假阳性的问题,提出了一种基于端到端的二维全卷积对象定位网络(2D FCN)与三维立体式目标分类卷积神经网络(3D CNN)相结合的肺结节检测方法。首先采用2D全卷积神经网络对所有CT图像进行初步检测,快速识别和定位CT图像中的疑似结节区域,输出一张与原图尺寸相同且被标记好的图像。然后计算疑似结节区域的坐标,根据坐标值提取疑似结节的三维立体图像块训练构建的3D卷积神经网络框架。最后利用训练的3D模型对候选结节做二分类处理以去除假阳性。在LIDC-IDRI数据集上,结节初步检测召回率在平均每位患者为36.2个假阳性时可达98.2%;在假阳性去除之后,假阳性为1和4时分别达到了87.3%和97.0%的准确率。LIDC-IDRI数据库上的实验结果表明,所提方法对三维CT图像的肺结节检测具有更高的适用性,取得了较高的召回率和准确率,优于目前相关文献报道的方法。该框架易于扩展到其他3D医疗图像的目标检测任务中,对辅助医师诊治具有重要的应用价值。

Abstract

Aiming at the problems that traditional lung nodules detection methods can only get low sensitivities and a lot of false positives, this paper presents a retrieval method for lung nodules CT image based on end-to-end two-dimensional full convolution object recognition network (2D FCN) and three-dimensional target classification convolution neural network (3D CNN). Firstly, the method builds the 2D CNN for candidate selection to detect and locate the suspected regions on axial slices, and outputs an image that is the same size as the original image and is marked. Secondly, the three-dimensional patches of each candidate are extracted to train the 3D CNN. Finally, the trained 3D model is used to classify the false positive nodules. Experimental results on the LIDC-IDRI dataset show that the proposed method can achieve the recall rate of nodules of 98.2% at 36.2 false positives per scan. In the false positive reduction, the method respectively achieves high detection sensitivities of 87.3% and 97.0% at 1 and 4 false positives per scan. Experimental results on the LIDC-IDRI dataset show that the proposed method is highly suited to be used for lung nodules detection, achieves high recall rate and accuracy and outperforms the current reported method. Meanwhile, the proposed framework is general and can be easily extended to many other 3D object detection tasks from volumetric medical images, and it has an important application value in clinical practice with the aid of radiologists and surgeons.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TP391

DOI:10.3788/lop55.051006

所属栏目:图像处理

基金项目:国家自然科学基金(61170120)

收稿日期:2017-11-02

修改稿日期:2017-11-23

网络出版日期:--

作者单位    点击查看

苗光:江南大学物联网工程学院, 江苏 无锡 214122
李朝锋:江南大学轻工过程先进控制教育部重点实验室, 江苏 无锡 214122

联系人作者:苗光(miao1094@126.com)

备注:苗光(1994-),男,硕士研究生,主要从事医学图像分割与病变检测方面的研究。E-mail: miao1094@126.com

【1】Sluimer I, Schilham A, Prokop M, et al. Computer analysis of computed tomography scans of the lung: a survey[J]. IEEE Transactions on Medical Imaging, 2006, 25(4): 385-405.

【2】Gao T, Gong J, Wang Y J, et al. Three dimensional adaptive template matching algorithm for lung nodule detection[J]. Journal of Image and Graphics, 2014, 19(9): 1384-1391.
高婷, 龚敬, 王远军, 等. 检测肺结节的3维自适应模板匹配[J]. 中国图象图形学报, 2014, 19(9): 1384-1391.

【3】Murphy K, Ginneken B V, Schilham A M R, et al. A large-scale evaluation of automatic pulmonary nodule detection in chest CT using local image features and k-nearest-neighbour classification[J]. Medical Image Analysis, 2009, 13(5): 757-770.

【4】Jacobs C, Rikxoort E M V, Twellmann T, et al. Automatic detection of subsolid pulmonary nodules in thoracic computed tomography images[J]. Medical Image Analysis, 2014, 18(2): 374-384.

【5】Chen S, Li L. A new computer aided diagnostic scheme for lung nodule detection on chest radiograph[J]. Acta Electronica Sinica, 2010, 38(5): 1211-1216.
陈胜, 李莉. 一种全新的基于胸片计算机辅助检测肺结节方案[J]. 电子学报, 2010, 38(5): 1211-1216.

【6】Wang M, Liu K X, Liu L, et al. Super-resolution reconstruction of image based on optimized convolution neural network[J]. Laser & Optoelectronics Progress, 2017, 54(11): 111005.
王民, 刘可心, 刘利, 等. 基于优化卷积神经网络的图像超分辨率重建[J]. 激光与光电子学进展, 2017, 54(11): 111005.

【7】Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6): 84-90.

【8】Ronneberger O, Fischer P, Brox T. U-Net: convolutional networks for biomedical image segmentation[C]. International Conference on Medical Image Computing and Computer-Assisted Intervention, [S.l.]: Springer, 2015: 234-241.

【9】Huang H, He K, Zheng X L, et al. Spatial-spectral feature extraction of hyperspectral image based on deep learning[J]. Laser & Optoelectronics Progress, 2017, 54(10): 101001.
黄鸿, 何凯, 郑新磊, 等. 基于深度学习的高光谱图像空-谱联合特征提取[J]. 激光与光电子学进展, 2017, 54(10): 101001.

【10】Chen H, Qi X, Yu L,et al. DCAN: deep contour-aware networks for accurate gland segmentation[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2016: 16541200.

【11】Kamnitsas K, Ledig C, Newcombe V F J, et al. Efficient multiscale 3D CNN with fully connected CRF for accurate brain lesion segmentation[J]. Medical Image Analysis, 2017, 36: 61-78.

【12】Setio A A A, Ciompi F, Litjens G, et al. Pulmonary nodule detection in CT images: false positive reduction using multi-view convolutional networks[J]. IEEE Transactions on Medical Imaging, 2016, 35(5): 1160-1169.

【13】van Ginneken B v, Setio A A A, Jacobs C, et al. Off-the-shelf convolutional neural network features for pulmonary nodule detection in computed tomography scans[C]. IEEE 12th International Symposium on Biomedical Imaging, 2015: 15309667.

【14】Ioffe S, Szegedy C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]. International Conference on Machine Learning, 2015: 448-456.

【15】He K, Zhang X, Ren S, et al. Delving deep into rectifiers: surpassing human-level performance on ImageNet classification[C]. IEEE International Conference on Computer Vision, 2015: 1026-1034.

【16】Glorot X, Bordes A, Bengio Y. Deep sparse rectifier neural networks[C]. 14th International Conference on Artificial Intelligence and Statistics, 2011: 315-323.

【17】Oliveira G L, Valada A, Bollen C, et al. Deep learning for human part discovery in images[C]. IEEE International Conference on Robotics and Automation, 2016: 16055460.

【18】Brown M S, Lo P, Goldin J G, et al. Toward clinically usable CAD for lung cancer screening with computed tomography[J]. European Radiology, 2014, 24(11): 2719-2728.

【19】Torres E L, Fiorina E, Pennazio F, et al. Large scale validation of the M5L lung CAD on heterogeneous CT datasets[J]. Medical Physics, 2015, 42(4): 1477-1489.

【20】Choi W J, Choi T S. Automated pulmonary nodule detection system in computed tomography images: a hierarchical block classification approach[J]. Entropy, 2013, 15(2): 507-523.

【21】Lii S G A, Mclennan G, Bidaut L, et al. The lung image database consortium(LIDC) and image database resource initiative (IDRI): a completed reference database of lung nodules on CT scans[J]. Medical Physics, 2011, 38(2): 915-931.

【22】van Ginneken B, Lii S G A, de Hoop B, et al. Comparing and combining algorithms for computer-aided detection of pulmonary nodules in computed tomography scans: the ANODE09 study[J]. Medical Image Analysis, 2010, 14(6): 707-722.

【23】Liu F, Liu P Y, Li B, et al. Deep learning model design of video target tracking based on TensorFlow platform[J]. Laser & Optoelectronics Progress, 2017, 54(9): 091501.
刘帆, 刘鹏远, 李兵, 等. TensorFlow平台下的视频目标跟踪深度学习模型设计[J]. 激光与光电子学进展, 2017, 54(9): 091501.

【24】Egan J P, Greenberg G Z, Schulman A I. Operating characteristics, signal detectability, and the method of free response[J]. Journal of the Acoustical Society of America, 1961, 33(8): 993-1007.

【25】Cascio D, Magro R, Fauci F, et al. Automatic detection of lung nodules in CT datasets based on stable 3D mass-spring models[J]. Computers in Biology & Medicine, 2012, 42(11): 1098-1109.

引用该论文

Miao Guang,Li Chaofeng. Detection of Pulmonary Nodules CT Images Combined with Two-Dimensional and Three-Dimensional Convolution Neural Networks[J]. Laser & Optoelectronics Progress, 2018, 55(5): 051006

苗光,李朝锋. 二维和三维卷积神经网络相结合的CT图像肺结节检测方法[J]. 激光与光电子学进展, 2018, 55(5): 051006

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF