肺癌是目前全球发病率和死亡率最高的癌症。肺结节是早期肺癌最主要、最常见的表现形式，准确检测肺结节对于前期肺癌的治疗有着十分关键的作用。目前医生主要靠CT扫描技术来诊断肺结节，但是由于肺部CT数量较多且肺癌早期的肺结节较小，医生仅靠肉眼进行分析与诊治，需要花费巨大的精力，而且极易出现漏诊或误诊等情况。针对目前存在的问题，本文研究了基于3D卷积神经网络的肺结节检测与分类算法，主要的研究内容如下：

（1）针对肺结节特征提取困难的问题，提出融合残差注意力网络的Faster R-CNN算法对候选结节进行检测。注意力机制可以通过抑制图像的无关特征将注意力集中在目标结节区域，提高特征提取能力。残差网络可以有效适应各种深度的网络，避免网络深度增加致使梯度退化的现象。通过结合残差注意力网络与Faster R-CNN，有效提高了肺结节的检测能力。实验结果表明该算法敏感度达到了96.5%，准确率达到了94.3%。

（2）为了进一步减少候选结节中假阳性的数量，提出了基于GC-ResNet的3D CNN算法。该算法融合了基于残差模块的GCNet，通过对全局上下文建模获取更多特征信息，提升了小尺寸结节的检测分类性能，达到减少假阳性的目的。此外引入DR Loss函数解决了训练数据类别不平衡的问题。实验结果表明该模型的CPM值达到了90.2%，验证了该算法在分类任务的有效性。

（3）结合本文提出的肺结节检测与分类算法，搭建了B/S架构的肺癌辅助检测系统。该系统实现了患者信息录入、患者病例查看、肺结节的自动检测等功能。最后对该系统进行了功能测试和性能测试。测试结果表明，该系统的各项功能及性能基本满足设计需求，对于临床肺结节检测具有一定的应用价值。

Lung cancer is currently the cancer with the highest incidence and mortality rate worldwide. Lung nodules are the most important and common manifestations of early lung cancer, and accurate detection of lung nodules plays a very critical role in the treatment of early lung cancer. Currently, doctors mainly rely on CT scan technology to diagnose lung nodules, but due to the large number of lung CTs and the small size of lung nodules in the early stage of lung cancer, doctors rely only on the naked eye to analyze and diagnose, which requires great efforts and is very prone to miss or misdiagnosis. To address the current problems, this paper investigates the lung nodule detection and classification algorithm based on 3D convolutional neural network, and the main research contents are as follows.

(1) Aiming at the difficulty of extracting pulmonary nodule features, this paper proposes a Faster R-CNN algorithm for the detection of candidate nodules by fusing residual attention networks. The attention mechanism can concentrate attention on the target nodule region by suppressing the unrelated features of the image, improve the feature extraction ability, and the residual network can effectively adapt to the network of various depths, avoiding the phenomenon of gradient degradation caused by the increase of network depth. By combining the residual attention network with Faster R-CNN, the detection ability of pulmonary nodules is effectively improved. Experimental results show that the sensitivity of the algorithm reaches 96.5% and the accuracy rate reaches 94.3%.

(2) In order to further reduce the number of false positives in candidate nodules, this paper proposes a GC-ResNet based 3D CNN algorithm. The algorithm integrates GCNet based on residual module, obtains more feature information by modeling the global context, improves the detection and classification performance of small size nodules, and achieves the purpose of reducing false positives. In addition, the Dr Loss function is introduced to solve the problem of unbalanced training data categories. Experimental results show that the CPM value of the model reaches 90.2%, which verifies the effectiveness of the algorithm in the classification task.

(3) Combining the lung nodule detection and classification algorithm proposed in this paper, an auxiliary lung cancer detection system with B/S architecture is built. The system realizes the functions of patient information entry, patient case view, and automatic detection of lung nodules. Finally, functional and performance tests are conducted on the system. The test results show that the functions and performance of the system basically meet the design requirements, and it has certain application value for clinical lung nodule detection.

[1]Sung H, Ferlay J, Siegel R L, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA: A Cancer Journal for Clinicians, 2021, 71(3): 209-249.

[2]Thandra K C, Barsouk A, Saginala K, et al. Epidemiology of lung cancer[J]. Contemporary Oncology, 2021, 25(1): 45.

[3]Mazzone P J, Lam L. Evaluating the patient with a pulmonary nodule: a review[J]. JAMA, 2022, 327(3): 264-273.

[4]Minna J D, Roth J A, Gazdar A F. Focus on lung cancer[J]. Cancer Cell, 2002, 1(1): 49-52.

[5]Gillies R J, Kinahan P E, Hricak H. Radiomics: images are more than pictures, they are data[J]. Radiology, 2016, 278(2): 563-577.

[6]Junior J R F, Koenigkam-Santos M, Cipriano F E G, et al. Radiomics-based features for pattern recognition of lung cancer histopathology and metastases[J]. Computer Methods and Programs in Biomedicine, 2018, 159: 23-30.

[7]Shin H C, Roth H R, Gao M, et al. Deep convolutional neural networks for computer-aided detection: CNN architectures, dataset characteristics and transfer learning[J]. IEEE Transactions on Medical Imaging, 2016, 35(5): 1285-1298.

[8]Rebecca L Siegel, Kimberly D Miller, and Ahmedin Jemal. Cancer statistics, 2019[J]. CA: A Cancer Journal for Clinicians, 2019, 69(1):7–34.

[9]Alberg A J, Samet J M. Epidemiology of lung cancer[J]. Chest, 2003, 123(1): 21S-49S.

[10]Firmino M, Morais A H, Mendoça R M, et al. Computer-aided detection system for lung cancer in computed tomography scans: review and future prospects[J]. Biomedical Engineering Online, 2014, 13(1): 1-16.

[11]Kim E S, Herbst R S, Wistuba I I, et al. The BATTLE trial: personalizing therapy for lung cancer[J]. Cancer Discovery, 2011, 1(1): 44-53.

[12]Shi Z, Li L, Suzuki K, et al. A New Computer Aided Detection System for Pulmonary Nodule Detection in Chest Radiography[J]. Advanced Science Letters, 2012, 11(1): 536-541.

[13]Pallis A G, Serfass L, Dziadziuszko R, et al. Targeted therapies in the treatment of advanced/metastatic NSCLC[J]. European Journal of Cancer, 2009, 45(14): 2473-2487.

[14]Shepherd F A, Douillard J Y, Blumenschein Jr G R. Immunotherapy for non-small cell lung cancer: novel approaches to improve patient outcome[J]. Journal of Thoracic Oncology, 2011, 6(10): 1763-1773.

[15]Kumar D, Wong A, Clausi D A. Lung nodule classification using deep features in CT images[C]//2015 12th Conference on Computer and Robot Vision. IEEE, 2015: 133-138.

[16]魏颖, 郭薇, 孙月芳, 等. 基于局部灰度最大和改进Mahalanobis距离分类的肺结节检测算法[J]. 中国图象图形学报, 2008, 2008(09): 1720-1726.

[17]Lee M C, Boroczky L, Sungur-Stasik K, et al. Computer-aided diagnosis of pulmonary nodules using a two-step approach for feature selection and classifier ensemble construction[J]. Artificial Intelligence in Medicine, 2010, 50(1): 43-53.

[18]Sun T, Wang J, Li X, et al. Comparative evaluation of support vector machines for computer aided diagnosis of lung cancer in CT based on a multi-dimensional data set[J]. Computer Methods and Programs in Biomedicine, 2013, 111(2): 519-524.

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

[20]Dhara A K, Mukhopadhyay S, Dutta A, et al. A combination of shape and texture features for classification of pulmonary nodules in lung CT images[J]. Journal of Digital Imaging, 2016, 29(4): 466-475.

[21]Wang J, Liu X, Dong D, et al. Prediction of malignant and benign of lung tumor using a quantitative radiomic method[C]//2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2016: 1272-1275.

[22]Wu W, Pierce L A, Zhang Y, et al. Comparison of prediction models with radiological semantic features and radiomics in lung cancer diagnosis of the pulmonary nodules: a case-control study[J]. European Radiology, 2019, 29(11): 6100-6108.

[23]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.

[24]Li W, Cao P, Zhao D, et al. Pulmonary nodule classification with deep convolutional neural networks on computed tomography images[J]. Computational and Mathematical Methods in Medicine, 2016, 2016: 1-7.

[25]Zhao X, Liu L, Qi S, et al. 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.

[26]Gupta A, Das S, Khurana T, et al. Prediction of lung cancer from low-resolution nodules in CT-scan images by using deep features[C]//2018 International Conference on Advances in Computing, Communications and Informatics (ICACCI). IEEE, 2018: 531-537.

[27]Sori W J, Feng J, Liu S. Multi-path convolutional neural network for lung cancer detection[J]. Multidimensional Systems and Signal Processing, 2019, 30(4): 1749-1768.

[28]Ai M, Lan B L, Chan W Y, et al. Lung nodule classification using deep local–global networks[J]. International Journal of Computer Assisted Radiology and Surgery, 2019, 14(10): 1815-1819.

[29]Ding J, Li A, Hu Z, et al. Accurate pulmonary nodule detection in computed tomography images using deep convolutional neural networks[C]//International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, Cham, 2017: 559-567.

[30]Huang X, Shan J, Vaidya V. Lung nodule detection in CT using 3D convolutional neural networks[C]//2017 IEEE 14th International Symposium on Biomedical Imaging (ISBI 2017). IEEE, 2017: 379-383.

[31]Hussein S, Cao K, Song Q, et al. Risk stratification of lung nodules using 3D CNN-based multi-task learning[C]//International Conference on Information Processing in Medical Imaging. Springer, Cham, 2017: 249-260.

[32]Liao F, Liang M, Li Z, et al. Evaluate the malignancy of pulmonary nodules using the 3-d deep leaky noisy-or network[J]. IEEE Transactions on Neural Networks and Learning Systems, 2019, 30(11): 3484-3495.

[33]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(05): 135-143.

[34]Zhu W, Liu C, Fan W, et al. Deeplung: Deep 3d dual path nets for automated pulmonary nodule detection and classification[C]//2018 IEEE Winter Conference on Applications of Computer Vision (WACV). IEEE, 2018: 673-681.

[35]Gong L, Jiang S, Yang Z, et al. Automated pulmonary nodule detection in CT images using 3D deep squeeze-and-excitation networks[J]. International Journal of Computer Assisted Radiology and Surgery, 2019, 14(11): 1969-1979.

[36]Li Y, Fan Y. DeepSEED: 3D squeeze-and-excitation encoder-decoder convolutional neural networks for pulmonary nodule detection[C]//2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI). IEEE, 2020: 1866-1869.

[37]Shi L, Ma H, Zhang J. Automatic detection of pulmonary nodules in CT images based on 3D res-i network[J]. The Visual Computer, 2021, 37(6): 1343-1356.

[38]Dou Q, Chen H, Yu L, et al. Automatic detection of cerebral microbleeds from MR images via 3D convolutional neural networks[J]. IEEE Transactions on Medical Imaging, 2016, 35(5): 1182-1195.

[39]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.

[40]Gu Y, Lu X, Yang L, et al. Automatic lung nodule detection using a 3D deep convolutional neural network combined with a multi-scale prediction strategy in chest CTs[J]. Computers in Biology and Medicine, 2018, 103: 220-231.

[41]高慧明, 赵涓涓, 刘继华, 等. 多尺度卷积神经网络用于肺结节假阳性降低[J]. 计算机工程与设计, 2019, 40(9): 2718-2724.

[42]吴保荣, 强彦, 王三虎, 等. 融合多维度卷积神经网络的肺结节分类方法[J]. 计算机工程与应用, 2019, 24:171-177.

[43]Yuan H, Fan Z, Wu Y, et al. An efficient multi-path 3D convolutional neural network for false-positive reduction of pulmonary nodule detection[J]. International Journal of Computer Assisted Radiology and Surgery, 2021: 1-9.

[44]Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2014: 580-587.

[45]He K, Zhang X, Ren S, 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.

[46]Girshick R. Fast r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision. 2015: 1440-1448.

[47]Ren S , He K , Girshick R , et al. Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 39(6):1137-1149.

[48]He K, Gkioxari G, Dollár P, et al. Mask r-cnn[C]//Proceedings of the IEEE International Conference on Computer Vision. 2017: 2961-2969.

[49]He K, Zhang X, Ren S, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016: 770-778.

[50]Woo S, Park J, Lee J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision (ECCV). 2018: 3-19.

[51]Hinton G E, Srivastava N, Krizhevsky A, et al. Improving neural networks by preventing co-adaptation of feature detectors[J]. ArXiv Preprint ArXiv:1207.0580, 2012, 3(4): 212-223.

[52]Lung nodule analysis 2016[EB/OL]. https://luna16.grand-challenge.org, 2016.

[53]Armato III S G, 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.

[54]Cao Y, Xu J, Lin S, et al. Gcnet: Non-local networks meet squeeze-excitation networks and beyond[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops. 2019. 1971-1980.