|本期目录/Table of Contents|

[1]张铭杰,李梦娜,程明瑶,等.高原肺水肿肺超声筛查深度学习模型构建与优化[J].中华肺部疾病杂志,2025,(01):68-73.[doi:10.3877/cma.j.issn.1674-6902.2025.01.011 ]
　Zhang Mingjie,Li Mengna,Chen Mingyao,et al.Construction and optimization of the deep learning model for pulmonary ultrasound screening with plateau pulmonary edema[J].,2025,(01):68-73.[doi:10.3877/cma.j.issn.1674-6902.2025.01.011 ]
点击复制

高原肺水肿肺超声筛查深度学习模型构建与优化(PDF)

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

卷:
期数:
2025年01期
页码:
68-73
栏目:
论著
出版日期:
2025-02-25

文章信息/Info

Title:
Construction and optimization of the deep learning model for pulmonary ultrasound screening with plateau pulmonary edema
作者:
张铭杰1; 李梦娜1; 程明瑶1; 王宇亮2; 粘永健3; 赵瑞臣2; 杨 宇1; 刘慕源1; 廖 元1; 唐 超1
850032 拉萨,西藏大学医学院1
850032 拉萨,西藏军区总医院重症医学科2
400038 重庆,陆军军医大学基础医学院3
Author(s):
Zhang Mingjie1;  Li Mengna1;  Chen Mingyao1;  Wang Yuliang2;  Nian Yongjian3;  Zhao Ruicheng1;  Yang Yu1;  Liu Muyuan1;  Liao Yuan1;  Tang Chao1.
1Medical School of Medicine, Xizang University, Lhasa 850032; 2Department of Critical Care Medicine,General Hospital of the Tibet Military Region, Lhasa 850032; 3School of Basic Medicine, Army Military Medical University,
关键词:
高原肺水肿;  肺部超声;  自动筛选技术;  模型;  人工智能
Keywords:
Deep learning;  High altitude pulmonary edema;  Pulmonary ultrasound;  Automatic screening technology;  Artificial intelligence
分类号:
R563
DOI:
10.3877/cma.j.issn.1674-6902.2025.01.011
摘要:
目的 利用西藏军区总医院收集的肺超声图像数据,基于深度学习技术,建立一种适用于高原肺水肿( high altitude pulmonary edema, HAPE)的肺超声筛查自动技术,用于自动筛选HAPE病症,提升诊断的准确性。方法 收集2021年1月至2023年12月期间,我院收治的高原肺水肿确诊病例共174例,按照分层的方式划分为训练集121例、验证集18例和测试集35例。采集患者肺超声图像,通过卷积神经网络(Convolutional Neural Network, CNN)模型进行图像识别和分析,系统进行多次训练和验证。模型被整合并优化以满足实时性和用户友好性需求,比较自动筛查技术系统与传统人工筛查方法诊断准确性。结果 模型性能评估中,AI模型的敏感性为95.00%,特异性为96.00%和总体准确率为95.50%(包含训练集115张,验证集17张,测试集33张),高于医师 组的敏感性84.33%,特异性87.67%和总体准确率85.50%(包含训练集106张,验证集16张,测试集31张)。统计学分析表明,AI系统与人工筛查方法在诊断敏感性、特异性及准确率上的差异具有统计学意义(P<0.05)。结论 与传统人工筛查方法相比,AI模型在诊断敏感性、特异性和准确率表现优异,可提高临床诊断。
Abstract:
Objective To establish an automatic lung ultrasound screening technology for high altitude pulmonary edema based on deep learning technology based on the ultrasonic lung image data collected by the General Hospital of Xizang Military Region, which can automatically screen the disease of HAPE and improve the diagnostic accuracy. Methods This study investigated the application effect of Convolutional Neural Network(CNN)based artificial intelligence(AI)model in the diagnosis of high altitude pulmonary edema. The study included 174 confirmed cases of high altitude pulmonary edema admitted to our hospital from January 2021 to December 2023. The cases were divided into 121 training sets, 18 verification sets and 35 test sets according to random stratification. The research methods include the collection of patients' lung ultrasound image data, the automatic recognition and analysis of the image using CNN model, and the training and verification of the model for several times to improve the diagnostic performance. In the model performance evaluation, the diagnostic accuracy, recall(sensitivity)and specificity of the AI system wTo establish an automatic lung ultrasound screening technology for high altitude pulmonary edema based on deep learning based on the data of lung ultrasound images collected by the General Hospital of Xizang Military Region, and to automatically screen HAPE and improve the diagnostic accuracy. Methods: This study investigated the application effect of Convolutional Neural Network(CNN)based artificial intelligence(AI)model in the diagnosis of high altitude pulmonary edema. The study included 174 confirmed cases of high altitude pulmonary edema admitted to our hospital from January 2020 to December 2023. The cases were divided into 121 training sets, 18 verification sets and 35 test sets according to random stratification. The research methods include the collection of patients'lung ultrasound image data, the automatic recognition and analysis of the image using CNN model, and the training and verification of the model for several times to improve the diagnostic performance. Result In the model performance evaluation, the sensitivity of the AI model was 95.00%, the specificity was 96.00%, and the overall accuracy rate was 95.50%(including 115 training set images, 17 validation set images, and 33 test set images), which was higher than the sensitivity of 84.33%, specificity of 87.67%, and overall accuracy rate of 85.50% of the physician group(including 106 training set images, 16 validation set images, and 31 test set images). Statistical analysis indicated that the differences in diagnostic sensitivity, specificity, and accuracy rate between the AI system and the manual screening method were statistically significant(P<0.05).Conclusion This study demonstrated the superior performance of CNN-based AI screening technology in the diagnosis of high altitude pulmonary edema. Compared with traditional manual screening methods, the AI model performs well in terms of diagnostic sensitivity, specificity and accuracy, and can effectively compensate for the limitations caused by the inexperience of doctors.

参考文献/References:

1 次仁曲珍, 卢海燕, 次旦旺久, 等. 高原地区肠系膜静脉血栓MSCT表现及临床特征分析[J]. 中国CT和MRI杂志, 2023, 21(05): 108-110.
2 王 健, 刘 辉, 李 成, 等. 低海拔地区官兵进驻高原后血液指标的变化[J]. 武警医学, 2020, 31(9): 762-764, 770.
3 刘小云, 孙 莉, 胡小兵. 部队进驻高原急性高原病预防策略分析[J]. 武警医学, 2021, 32(11): 995-998.
4 马德花, 鲍海咏. 非诺贝特联合易善复治疗高原地区非酒精性脂肪肝疗效观察[J]. 高原医学杂志, 2017, 27(2): 20-22.
5 Howell L, Ingram N, Lapham R, et al. Deep learning for real-time multi-class segmentation of artefacts in lung ultrasound[J]. ULTRASONICS, 2024, 140: 107251.
6 Damjan V, Andrew W, Maria A, et al. Automatic deep learning-based pleural effusion segmentation in lung ultrasound images[J]. BMC Med Inform Decis Mak, 2023, 23(1): 274.
7 周飞燕, 金林鹏, 董 军. 卷积神经网络研究综述[J]. 计算机学报, 2017, 40(6): 1229-1251.
8 周志华, 陈世福. 神经网络集成[J]. 计算机学报, 2002, 25(1): 1-8.
9 巩 高, 黄文华, 曹 石, 等. 人工智能在医学的应用研究进展[J]. 中国医学物理学杂志, 2021, 38(8): 1044-1047.
10 Lucio CL, Federico M, Francesco T, et al. Multi-objective automatic analysis of lung ultrasound data from COVID-19 patients by means of deep learning and decision trees[J]. Appl Soft Comput, 2023, 133: 109926
11 吴冠楠, 陈 晨, 顾晓凌, 等.间质性肺疾病的影像学评估: 肺部超声能做什么?[J]. 中华结核和呼吸杂志, 2024, 47(2): 172-177.
12 赵浩天, 刘 奕, 刘元琳, 等. 对比床旁急诊肺超声与联合心肺及附加超声诊断急性呼吸困难病因的准确性[J]. 中国介入影像与治疗学, 2024, 21(3): 134-138.
13 袁 灵, 钟静怡, 江 兵, 等. 以肺部超声为应用的多学科诊疗体系在高原性肺水肿诊疗中的价值[J]. 成都医学院学报, 2024, 19(1): 94-97.
14 邬晓臣, 王 舰, 刘 建, 等. 高原地区婴幼儿完全型房室间隔缺损的外科处理策略及随访结果[J]. 心脏杂志, 2024, 36(1): 50-53.
15 戴礼鸣, 凌祥伟, 李 维, 等. 超声评估肺不张在全麻中的应用[J]. 中国医师杂志, 2024, 26(4): 564-567.
16 刘 敬. 超声监测下管理肺疾病:或使早产儿支气管肺发育不良成为可避免的疾病[J]. 中国当代儿科杂志, 2024, 26(1): 14-18.
17 丁雅杰, 孙医学, 李 阳, 等. 超声造影对不同大小肺周围型病变活检的指导价值[J]. 中华全科医学, 2024, 22(2): 217-221.
18 蔡书静, 张乐乐, 陈思悦, 等. 肺部超声在儿童社区获得性肺炎中的诊断价值研究[J]. 中华儿科杂志, 2024, 62(4): 331-336.
19 封在李, 杨明杰, 雷 嘉, 等. 肺脏超声在诊断NRDS及规范指导肺表面活性物质治疗中的应用研究[J]. 重庆医学, 2024, 53(3): 339-343.
20 刘 珂, 张婧娴, 王如刚. 肺超声纹理特征ARDS与心源性肺水肿的鉴别诊断意义[J/CD]. 中华肺部疾病杂志(电子版), 2023, 16(6): 892-894.
21 吴赤球, 林海平, 余 丹, 等. 三种超声心动图经验公式评估高原某部官兵肺动脉压的比较[J]. 武警医学, 2023, 34(1): 1-4.
22 范浩浩, 姜倩倩, 邢文宇, 等. 基于人工智能的自动肺部超声评分对ARDS患者血管外肺水评估的价值[J]. 中国急救医学, 2023, 43(1): 24-29.
23 张清友, 马宇德, 高雨萌, 等. 三尖瓣环收缩期位移在婴幼儿高原性心脏病右心功能评估中的应用[J]. 中国实用儿科杂志, 2022, 36(5): 347-351.
24 薛 寒, 陈小枫, 缪小莉, 等. 床旁肺部超声用于心脏术后机械通气患者的物理治疗[J]. 生物医学工程与临床, 2022, 26(1): 29-34.
25 张晓花, 王 锟, 张 平, 等. 肺动脉分支起源异常的产前超声筛查与诊断[J]. 中国临床医学影像杂志, 2022, 33(7): 485-488.
26 沈 珀, 沈亚南, 张 晨, 等. 肺超声评分评价肺保护性通气策略对老年开腹手术患者肺损伤的影响[J]. 临床麻醉学杂志, 2021, 37(9): 901-905.
27 田铭君, 黄立伟, 郑敏娟. 单侧肺动脉缺如的临床及超声影像特征分析[J]. 中国超声医学杂志, 2021, 37(6): 628-631.
28 郑高原, 佘世刚. 基于物联网的高精度超声波气体流量监测系统设计[J]. 仪表技术与传感器, 2021, 57(2): 65-70.
29 朱永城, 江慧琳, 伍卓文, 等. 肺部超声对急诊尿毒症性急性肺水肿患者无创正压通气的疗效评估[J]. 中国急救医学, 2020, 40(8): 715-718.
30 任婷婷, 李鸿斌. 超声在结缔组织病相关肺间质病变中的应用[J]. 中国实用内科杂志, 2020, 40(2): 166-169, 173.
31 刘 敬, 冯 星, 胡才宝, 等. 新生儿肺脏疾病超声诊断指南[J]. 中国当代儿科杂志, 2019, 21(2): 105-113.
32 赵 旭, 薛林燕, 白国银, 等. 二维超声对高原地区正常胎儿生长规律初探[J]. 中国医学计算机成像杂志, 2019, 25(1): 72-75.

备注/Memo

备注/Memo:
基金项目: 中国博士后科学基金(2017M623356)
通信作者: 王宇亮, Email: wangtmmu@163.com
粘永健, Email: yjnian@126.com
更新日期/Last Update: 2025-02-20