北京生物医学工程

基于非结节自动分类的二维卷积网络在肺结节检测假阳性减少中的应用

The two-dimensional convolution network based on non nodule automatic classification for reduction of false positivity in pulmonary nodule detection

作者：任敬谋李晓琴

单位：北京工业大学环境与生命学部(北京 100124)

关键词：医学影像处理; 计算机断层扫描; 肺结节检测; 卷积神经网络; 深度学习

分类号：R318.04

出版年·卷·期（页码）：2020·39·4（389-397）

摘要：

目的针对计算机断层扫描(computed tomography, CT)图像的肺结节自动检测中灵敏度低及存在大量假阳性的问题,本文提出了一种基于非结节自动分类的二维卷积神经网络( convolutional neural network, CNN),并用于肺结节检测中的假阳性减少。方法首先对 CT 图像进行预处理,通过对原始 CT 图像重采样和归一化,解决不同样本像素间隔不一致及图像对比度不统一问题;采用结节不同空间方向的二维切片信息采集进行正样本扩充,负样本无监督分类方法平衡正负样本数量;分别利用不同类别负样本与正样本训练二维卷积神经网络,获得多个用于降低假阳性的 2D CNN 肺结节检测模型,对LUNA16 提供的假阳性减少数据集进行五折交叉验证,利用官方提供的评估程序对模型进行评估。结果通过与直接使用单个 2D CNN 进行分类的模型比较,对非结节分类后训练多个模型的分类结果较佳,最终竞争性指标(competition performance metric,CPM)竞争性得分 0.849? 结论基于非结节自动分类的 2D CNN 模型可以有效地对假阳性肺结节进行剔除,相较于其他 2D CNN 具有竞争力,可为肺癌早期筛查提供帮助。

Objective In order to solve the problem of low sensitivity and a large number of false positives in the automatic detection of pulmonary nodules in CT images, this paper proposes a two-dimensional convolutional neural network ( CNN ) based on non-nodule automatic classification and applies it to the reduction of false positives in the detection of pulmonary nodules. Methods Firstly, the CT image was preprocessed by resampling and normalizing the original CT image, to solve the problem of inconsistent pixel spacing and image contrast of different samples. The positive samples were expanded by two-dimensional slice information collection in different spatial directions, and the negative samples were classified unsupervised to balance the positive and negative samples. Two-dimensional convolutional neural networks were trained with different types of negative samples and positive samples to obtain a number of 2D CNN pulmonary nodule detection models for reducing false-positive, by using the false positive reduction data set provided by LUNA16 to conduct 5-fold cross validation, and evaluated the model with the evaluation procedure provided by LUNA16. Results Compared with the model directly using a single 2D CNN for classification, the result of training multiple models after non nodule classification was better, the final CPM ( competition performance metric) competitive score was 0.849. Conclusions The 2D CNN model based on non-nodule automatic classification can effectively reduce the number of false-positive pulmonary nodules, which is competitive with other 2D CNN, and can provide help for early lung cancer screening.

参考文献：

[ 1 ] Fitzmaurice C, Allen C, Barber R M, et al. Global, regional, and national cancer incidence, mortality, years of life lost, years lived with disability, and disability-adjusted life-years for 32 cancer groups, 1990 to 2015:a systematic analysis for the Global Burden of Disease Study[J]. JAMA Oncology, 2017, 3(4):524-548.

[ 2 ] Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018:GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[ J]. CA:A Cancer Journal for Clinicians, 2018, 68(6):394-424.

[ 3 ] Ferlay J, Colombet M, Soerjomataram I, et al. Cancer incidence and mortality patterns in Europe:Estimates for 40 countries and 25 major cancers in 2018 [ J ] . European Journal of Cancer, 2018, 103:356-387.

[ 4 ] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018 [ J ] . CA:A Cancer Journal for Clinicians, 2018, 68(1) :7-30.

[ 5 ] Ding L, Getz G, Wheeler DA, et al. Somatic mutations affect key pathways in lung adenocarcinoma [ J] . Nature, 2008, 455 (7216) :1069-1075.

[ 6 ] Sanders HR, Albitar M. Somatic mutations of signaling genes in non-small-cell lung cancer [ J ] . Cancer Genetics and Cytogenetics, 2010, 203(1) :7-15.

[ 7 ] Jemal A, Center MM, DeSantis C, et al. Global patterns of cancer incidence and mortality rates and trends[J]. Cancer Epidemiology Biomarkers & Prevention, 2010, 19(8):1893-1907.

[ 8 ] Oken MM, Hocking WG, Kvale PA, et al. Screening by chest radiograph and lung cancer mortality the prostate, lung, colorectal, and ovarian ( PLCO) randomized trial [ J] . JAMA- Journal of the American Medical Association, 2011, 306( 17) : 1865-1873.

[ 9 ] National Lung Screening Trial Research Team. The national lung screening trial:overview and study design[ J] . Radiology, 2011, 258(1) :243-253.

[10] Aberle DR, Adams AM, Berg CD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening[ J] . the New England Journal of Medicine, 2011, 365(5) :395-409.

[11] Cao P, Liu X, Yang J, et al. A multi-kernel based framework for heterogeneous feature selection and over-sampling for computer-aided detection of pulmonary nodules [ J ] . Pattern Recognition, 2017, 64:327-346.

[12] Shen W, Zhou M, Yang F, et al. Multi-crop convolutional neural networks for lung nodule malignancy suspiciousness classification[ J] . Pattern Recognition, 2017,61:663-673.

[13] Liu X, Hou F, Qin H, et al. Multi-view multi-scale CNNs for lung nodule type classification from CT images [ J ] . Pattern Recognition, 2018, 77:262-275.

[14] Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020 [ J ] . CA:A Cancer Journal for Clinicians, 2020, 70(1) :7-30.

[15] Ye X, Lin X, Dehmeshki J, et al. Shape-based computer-aided detection of lung nodules in thoracic CT images [ J ]. IEEE Transactions on Biomedical Engineering, 2009, 56 ( 7 ): 1810-1820.

[16] Jacobs C, van Rikxoort EM, Twellmann T, et al. Automatic detection of subsolid pulmonary nodules in thoracic computed tomography images[ J]. Medical Image Analysis, 2014, 18( 2): 374-384.

[17] Ali I, Hart GR, Gunabushanam G, et al. Lung nodule detection via deep reinforcement learning [ J ] . Frontiers in Oncology, 2018, 8:108.

[18] Dou Q, Chen H, Yu L, et al. Multilevel contextual 3-D CNNs for false positive reduction in pulmonary nodule detection [ J] .IEEE Transactions on Biomedical Engineering, 2017, 64 ( 7) :1558-1567.

[19] Ciompi F, de Hoop B, van Riel SJ, et al. 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.

[20] Setio AAA, 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.

[21] Armato Ⅲ SG, 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] Setio AAA, Traverso A, de Bel T, et al. Validation, comparison,and combination of algorithms for automatic detection of pulmonary nodules in computed tomography images:The LUNA16 challenge [J]. Medical Image Analysis, 2017, 42:1-13.

[23] Murphy K, van Ginneken B, Schilham A, 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.

[24] Setio AAA, Jacobs C, Gelderblom J, et al. Automatic detection of large pulmonary solid nodules in thoracic CT images [ J ] . Medical Physics, 2015, 42(10) :5642-5653.

[25] Ponti MA, Ribeiro LSF, Nazare TS, et al. Everything you wanted to know about Deep Learning for Computer Vision but were afraid to ask [ M] / / 2017 30th SIBGRAPI Conference on Graphics, Patterns and Images Tutorials ( SIBGRAPI - T ) .Niterói, Brazil:IEEE Press, 2017:17-41.

[26] 吕晓琪, 吴凉, 谷宇, 等. 基于三维卷积神经网络的低剂量CT肺结节检测[J]. 光学精密工程, 2018, 26(5):1211-1218.

Lyu XQ, Wu L, GU Y, et al. Detection of low dose CT pulmonary nodules based on 3D convolution neural network[ J] . Optics and Precision Engineering, 2018, 26(5) :1211-1218.

[27] 金奇樑.基于 CT 图像的肺结节自动识别系统研究[ D] .杭州:浙江大学, 2016:40-50.

Jin QL. Research on pulmonary nodules detection system based on CT images[ D] . Hangzhou:Zhejiang University, 2016:40-50.

[28] Dandil E, ?akirogˇ lu M, Ek?i Z, et al. Artificial neural network-based classification system for lung nodules on computed tomography scans [ C ] / / International Conference of Soft Computing and Pattern Recognition ( SoCPaR) . Tunis, Tunisia:IEEE Press, 2014:382-386.

服务与反馈：

【文章下载】【加入收藏】

提示：您还未登录，请登录！点此登录