«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《中华肺部疾病杂志》[ISSN:1006-6977/CN:61-1281/TN]

卷:

期数:

2024年05期

页码:

673-679

栏目:

论著

出版日期:

2024-10-25

文章信息/Info

Title:

Establishment of a prediction model for benign and malignant pulmonary nodules based on 3D Res U-Net-Faster RCNN and CT imaging features

作者:

犹成亿¹; 2; 尤 恒¹; 2; 叶东樊¹; 2; 张 雯¹; 2; 刘 禹¹; 2; 王仁宇³; 苏琳茜⁴; 甘 慧⁵; 徐 智¹; 2

400037 重庆,陆军(第三)军医大学第二附属医院呼吸与危重症医学中心¹
400037 重庆,重大呼吸疾病精准诊疗与防控重庆市重点实验室²
400037 重庆,陆军(第三)军医大学第二附属医院胸外科³
400037 重庆,陆军(第三)军医大学第二附属医院病理科⁴
400037 重庆,陆军(第三)军医大学第二附属医院放射科⁵

Author(s):

You Chengyi¹; 2;  You Heng¹; 2;  Ye Dongfang¹; 2;  Zhang Wen¹; 2;  Liu Yu¹; 2;  Wang Renyu³;  Su Linxi⁴;  Gan Hui⁵;  Xu Zhi¹; 2.

¹Department of Respiratory and Critical Care Medical Center, The Second Affiliated Hospital of Army Medical University, Chongqing 400037, China; ²Chongqing Key Laboratory for precise diagnosis, treatment, prevention and control of major respiratory diseases,Chongqing 400037, China; ³Department of Thoracic Surgery, the Second Affiliated Hospital of Army Medical University, Chongqing 400037, China; ⁴Department of Pathology, the Second Affiliated Hospital of Army Medical University, Chongqing 400037, China; ⁵Department of Radiology, the Second Affiliated Hospital of Army Medical University, Chongqing 400037, China

关键词:

肺结节;  3D残差U型网络;  基于区域的卷积神经网络;  预测模型

Keywords:

Pulmonary nodules;  3D residual u-net;  Faster region-based convolutional neural networks;  Predictive model

分类号:

R563

DOI:

10.3877/cma.j.issn.1674-6902.2024.05.001

摘要:

目的 基于3D Res U-Net-Faster RCNN技术采用组织病理学确诊的肺结节及胸部CT影像学数据,构建肺结节性质的预测模型。方法 选择2020年10月至2023年10月我院收治的经外科手术切除并具有组织病理学诊断肺结节患者528例,其中恶性肺结节442例,良性肺结节86例,按7︰3随机分为训练集和测试集。采用改良的3D残差U型网络(3D residual u-net, 3D Res U-Net)融合基于区域的卷积神经网络(faster region-based convolutional neural networks, faster RCNN)模型,通过识别肺结节感兴趣区域(region of interest, ROI)和提取CT影像学特征,构建肺结节性质预测模型,筛选判断恶性肺结节CT影像特征权重。通过混淆矩阵、精准率、召回率、F1值、Dice相似系数(dice loss)、受试者工作特征曲线(receiver operating characteristic, ROC)判断该模型对肺结节性质的诊断精准度,采用外部数据验证模型工作性能。结果 基于3D Res U-Net-Faster RCNN技术构建的肺结节性质预测模型,分割ROI的Dice Loss为0.85,测试集对恶性肺结节识别的精确度为0.85,召回率0.76,F1值0.80,曲线下面积(area under the curve, AUC)值0.86。对外部验证集的肺结节识别准确率0.86,恶性结节识别精确度0.92,召回率0.87,F1值0.90; 良性结节识别精确度0.92,召回率0.82,F1值0.87。肺CT 影像特征中平均灰度值、最大直径与体积比值、表面积与体积比值对恶性肺结节预测的权重高。良、恶性结节组间的肺结节直径大小、毛刺征、血管穿行征具有显著差异(P<0.05)。结论 基于3D Res U-Net-Faster RCNN技术对CT影像学特征构建的人工智能(artificial intelligence, AI)驱动诊断模型对肺结节性质具有预测性能,对提高早期肺癌的筛查具有临床诊断意义。

Abstract:

Objective The 3D Res U-Net-Faster RCNN model was meticulously trained and learned using CT imaging data of histopathology-confirmed pulmonary nodules to develop a predictive model capable of distinguishing between benign and malignant pulmonary nodules. Methods A retrospective study incorporating 528 pulmonary nodules cases(malignant 442 and benign 86 )treated between October 2020 and October 2023 was conducted. The cases were randomly divided into a training set and a test set in a 7︰3 ratio. A modified 3D residual U-shaped network, fused with faster region-based convolutional neural networks(Faster R-CNN), was utilized to identify regions of interest(ROI)within pulmonary nodules and to extract CT imaging features. This approach was employed to construct a predictive model capable of distinguishing between benign and malignant pulmonary nodules, as well as to judge and screen the CT imaging feature weights of malignant nodules. The diagnostic accuracy of the model for pulmonary nodules was determined by confusion matrix, accuracy, by precision, recall, F1 value, Dice similarity coefficient(dice loss), subject operating characteristic curve(receiver operating characteristic, ROC), and the performance of the model was verified with external data. Results In the pulmonary nodule properties prediction model based on 3D Res U-Net-Faster RCNN deep learning technology, the Dice Loss of segmentation ROI was 0.85, and the accuracy of the test set to identify malignant lung nodules was 0.85, the recall rate 0.76, F1 value 0.80, and the area under the curve(area under the curve, AUC)value 0.86. For the external validation set, lung nodules identification accuracy was 0.86, malignant nodules identification accuracy 0.92, recall rate 0.87, F1 value 0.90; benign nodules identification accuracy 0.92, recall rate 0.82, and F1 value 0.87. The mean gray value, maximum diameter to volume ratio, and surface area to volume ratio gave the highest weight to the prediction of malignant lung nodules. The diameter, burr and vascular travel were significantly different between the benign and malignant nodules(P<0.05). Conclusion The AI-driven diagnostic model constructed based on the CT-imaging features of deep learning of 3D Res U-Net-Faster RCNN technology has good predictive performance for the properties of pulmonary nodules and has auxiliary diagnostic significance for improving the screening of early lung cancer.

参考文献/References:

1 Global cancer burden growing, amidst mounting need for services.[J]. Saudi Med J, 2024, 45(3): 326-327.
2 Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022[J]. J Nat Cancer Center, 2024, 4(1): 47-53.
3 Changing profiles of cancer burden worldwide and in China: a secondary analysis of the global cancer statistics 2020[J]. Chin Med J, 2021, 134(7): 783-791.
4 《早期肺癌诊断中国专家共识(2023年版)》发布[J]. 中华医学信息导报, 2023, 38(3): 10-11.
5 Xu K, Lv Y, Chen T, et al. Evaluation of the novel international association for the study of lung cancer grading system in adenocarcinoma with spread through air space.[J]. Translat Lung Cancer Res, 2024, 13(8): 1862-1876.
6 田舜彧, 李晓雪. 低剂量螺旋CT筛查肺癌及肺结节的研究进展[J]. 医疗装备, 2024, 37(14): 161-164.
7 中华医学会肿瘤学分会. 中华医学会肺癌临床诊疗指南(2024版)[J]. 中华肿瘤杂志, 2024, 46(9): 805-843.
8 中华医学会肿瘤学分会, 中华医学会杂志社. 中华医学会肺癌临床诊疗指南(2022版)[J]. 中华肿瘤杂志, 2022, 44(6): 457-490.
9 Sarah H, Pranav J, Yousaf H, et al. Utility of FDG PET/CT for assessment of lung nodules identified during low dose computed tomography screening.[J]. BMC Med Imag, 2020, 20(1): 69-70.
10 杨雪玲, 于海鹏, 司同国. 胸部肿瘤经皮穿刺活检中国专家共识(2020版)[J]. 中华介入放射学电子杂志, 2021, 9(2): 117-126.
11 刘 娜, 赵正凯, 邹佳瑜, 等. 基于人工智能的胸部CT肺结节检出及良恶性诊断效能评估[J]. CT理论与应用研究, 2021, 30(6): 709-715.
12 Beibei J, Nianyun L, Xiaomeng S, et al. Deep learning reconstruction shows better lung nodule detection for ultra-low-dose chest ct.[J]. Radiology, 2022, 303(1): 202-212.
13 Shubham D, Annappa B, A. M P. Recent advancements in deep learning based lung cancer detection: A systematic review[J]. Engineer Applicat Artif Intell, 2022, 116(105490): 1-15.
14 Akitoshi S, Daiju U, Antoine C, et al. Deep learning-based algorithm for lung cancer detection on chest radiographs using the segmentation method[J]. Scientific Reports, 2022, 12(1): 1-11.
15 Ueda D, Shimazaki A, Miki Y. Technical and clinical overview of deep learning in radiology[J]. Jpn J Radiol, 2019, 37(1): 15-33.
16 Ou WC, Polat D, Dogan BE. Deep learning in breast radiology: current progress and future directions[J]. Eur Radiol, 2021, 31(7): 4872-4885.
17 Saba L, Biswas M, Kuppili V, et al. The present and future of deep learning in radiology[J]. Eur J Radiol, 2019, 114: 14-24.
18 刘 莹, 杨 硕. 基于改进UNet网络的室内运动目标阴影分割[J]. 计算机系统应用, 2022, 31(12): 412-419.
19 蒋文斌, 彭 晶, 叶阁焰. 深度学习自适应学习率算法研究[J]. 华中科技大学学报(自然科学版), 2019, 47(5): 79-83.
20 Yang L, Qianqian Y, Haitao Y, et al. Automated segmentation of vertebral cortex with 3D U-Net-based deep convolutional neural network[J]. Front Bioengineer Biotechnol, 2022, 10: 1-13.
21 Jing X, Haojie R, Shenzhou C, et al. An improved faster R-CNN algorithm for assisted detection of lung nodules[J]. Comput Biol Med, 2022, 153: 1-12.
22 Hui Z, Yanjun P, Yanfei G. Pulmonary nodules detection based on multi-scale attention networks[J]. Scientific Reports, 2022, 12(1): 1-14.
23 Yue Z, Zhongyang W, Xinyao L, et al. Pulmonary Nodule Detection Based on Multiscale Feature Fusion[J]. Computat Mathemat Meth Med, 2022, 2022: 1-11.
24 Hou X, Wu M, Chen J, et al. Establishment and verification of a prediction model based on clinical characteristics and computed tomography radiomics parameters for distinguishing benign and malignant pulmonary nodules.[J]. J Thorac Dis, 2024, 16(3): 1-12.
25 Minmini S, Anupama C, Abjasree S, et al. Radiomics as a non-invasive adjunct to Chest CT in distinguishing benign and malignant lung nodules[J]. Scient Reports, 2023, 13(1): 1-14.
26 Shi Z, Zhang X, Jiang T. Study progress of radiomics in precision medicine for lung cancer[J]. Chin J Lung Cancer, 2019, 22(6): 933-939.
27 王 璐, 洪群英. 肺结节诊治中国专家共识(2018年版)解读[J]. 中国实用内科杂志, 2019, 39(05): 440-442.
28 Daungsupawong H, Wiwanitkit V. Radiomics and clinical data for the diagnosis of incidental pulmonary nodules and lung cancer screening: correspondence[J]. Archivos de Bronconeumol, 2024, 60(10): 1-3.
29 Hosny A, Parmar C, Quackenbush J, et al. Artificial intelligence in radiology[J]. Nature Reviews Cancer, 2018, 18(8): 500-510.
30 Coelho LP. How artificial intelligence is shaping medical imaging technology: a survey of innovations and applications[J]. Bioengineering, 2023, 10(12): 1-21.
31 吴绯红, 赵煌旋, 杨 帆, 等. 医学影像+人工智能的发展、现状与未来[J]. 临床放射学杂志, 2022, 41(4): 764-767.
32 倪 炯, 王培军. 医学影像人工智能的现状与未来[J]. 中华医学杂志, 2021, 101(7): 455-457.
33 范卫杰, 张 冬. 影像组学及深度学习在肺结节良恶性鉴别诊断中的新理念[J/CD]. 中华肺部疾病杂志(电子版), 2021, 14(5): 549-553.
34 曹 宇, 邢素霞, 逄键梁, 等. 基于改进的VGG-16卷积神经网络的肺结节检测[J]. 中国医学物理学杂志, 2020, 37(7): 940-944.
35 Hui Y, Jinqiu L, Lixin Z, et al. Design of lung nodules segmentation and recognition algorithm based on deep learning[J]. BMC Bioinformatics, 2021, 22(16): 1-12.
36 杨 愉, 谭雨豪, 王丽嘉. 改进的机器学习模型在肺结节良恶性分类中的研究[J]. 计算机与数字工程, 2024, 52(7): 2227-2232.

备注/Memo

备注/Memo:

基金项目: 重庆市技术创新与应用发展专项重点项目(CSTC2021jscx-gksb-N0029)
通信作者: 徐 智, Email: xuzhihxk@tmmu.edu.cn

常用功能

更新日期/Last Update: 2024-10-25