基于深度学习的人工智能测量婴儿非偏心型髋关节的研究

doi:10.3877/cma.j.issn.1672-6448.2021.05.006

中华医学超声杂志（电子版） ›› 2021, Vol. 18 ›› Issue (05) : 467 -471. doi: 10.3877/cma.j.issn.1672-6448.2021.05.006

所属专题：文献；

小儿超声影像学

基于深度学习的人工智能测量婴儿非偏心型髋关节的研究

郭爽萍¹, 倪东², 尚宁¹^,^†(

), 王丽敏¹, 胡歆迪³, 吕栩再⁴, 梁永栋¹

^1. 511400　广州，广东省妇幼保健院超声诊断科
^2. 511400　广州，广东省妇幼保健院医务科
^3. 518071　深圳大学医学部生物医学工程学院
^4. 211166　南京医科大学生物医学工程与信息学院

收稿日期:2020-10-19 出版日期:2021-05-01
通信作者: 尚宁

Use of deep learning-based artificial intelligence to measure the centred hip joint of infants

Shuangping Guo¹, Dong Ni², Ning Shang¹(), Limin Wang¹, Xindi Hu³, Xuzai Lyu⁴, Yongdong Liang¹

^1. Department of Ultrasound, Guangdong Women and Children's Hospital, Guangzhou 511400, China
^2. Shenzhen University Health Science Center School of Biomedical Engineering, Shenzhen 518071, China
^3. the School of Biomedical Engineering and Information, Nanjing Medical University, Nanjing 211166, China
^4. Medical Department, Guangdong Women and Children's Hospital, Guangzhou 511400, China

Received:2020-10-19 Published:2021-05-01
Corresponding author: Ning Shang

全文 ( ) 下载PDF ( ) 导出引用

目的

评估人工智能模型自动测量婴儿髋关节发育指标的临床应用价值。

方法

选取2019年1月至11月于广东省妇幼保健院超声科进行髋关节超声检查的婴儿231例（共462张标准髋关节图片），使用组内相关系数（ICC）、Deming回归对高年资医师与人工智能测量髋关节数据进行一致性分析，使用ICC对高年资医师、低年资医师、人工智能测量数据进行观察者间一致性分析，对高年资与低年资医师测量数据进行观察者内一致性分析。

结果

高年资医师与人工智能测量α角和β角的ICC分别为0.823、0.745；Deming回归斜率分别为0.856、1.205；回归残差图围绕参考线相对对称。高年资医师与低年资医师测量α角的ICC为0.77，β角的ICC为0.70；低年资医师与人工智能测量α角的ICC为0.79，β角的ICC为0.71；高年资医师与人工智能测量α角的ICC为0.87，β角的ICC为0.79。高年资医师测量2次α角的ICC为0.89，β角的ICC为0.84；低年资医师测量2次α角的ICC为0.88，β角的ICC为0.82。

结论

深度学习的人工智能模型测量髋关节数据与医师测量数据一致性好，且更接近高年资医师测量水平，可辅助超声医师进行临床早期筛查和诊断。

关键词: 发育性髋关节发育不良, 超声, 人工智能, 深度学习

Objective

To evaluate the clinical value of artificial intelligence model for automatic measurement of infant hip joint development.

Methods

A total of 231 infants who underwent hip joint ultrasonic examination (with 462 standard hip joint images) at the Ultrasound Diagnosis Department of Guangdong Women and Children Hospital from January to November 2019 were selected. Intraclass correlation coefficient (ICC) and Deming regression were used for consistency analysis of measurement data of hip joint between a senior sonographer and artificial intelligence. ICC was used for consistency analysis of measurement data between a senior sonographer, a junior sonographer, and artificial intelligence, and for intra-observer agreement of the senior sonographer and junior sonographer.

Results

The ICCs of α angle and β angle measured by the senior sonographer and artificial intelligence were 0.823 and 0.745, respectively; Deming regression slopes were 0.856 and 1.205, respectively, and regression residual plots were relatively symmetric. The ICCs of α angle and β angle measured by the senior sonographer and junior sonographer were 0.77 and 0.70, the ICCs of α angle and β angle measured by the junior sonographer and artificial intelligence were 0.79 and 0.71, and the ICCs of α angle and β angle measured by the senior sonographer and artificial intelligence was 0.87 and 0.79. The ICC of α angle double measured by the senior sonographer was 0.89, and the ICC of β angle was 0.84. The ICC of α angle double measured by the junior sonographer was 0.88, and the ICC of β angle was 0.82.

Conclusion

There is good data consistency between the deep learning-based artificial intelligence model and sonographers in measuring infant hip joints. The accuracy of artificial intelligence is closer to that of senior sonographer, and it can be used to assist in clinical ultrasound screening and diagnosis.

Key words: Developmental dysplasia of the hip, Ultrasound, Artificial intelligence, Deep learning

图1 髋关节超声检查图像的示意图。图a为人工标记的髋关节切面图；图b显示4个关键解剖结构；图c显示3个标记点；图d为α角和β角的测量示意图

图2 人工智能模型自动化测量方法流程图

表1 高年资医师与人工智能测量髋关节α角和β角结果比较（°，

x ˉ

±s）

测定方法	例数	α角	β角
人工智能模型	231	65.63±4.39	66.11±5.51
高年资医师	231	66.04±3.86	66.03±6.33
t值		3.66	0.435
P值		＜0.001	0.664

表1 高年资医师与人工智能测量髋关节α角和β角结果比较（°，

x ˉ

±s）

测定方法	例数	α角	β角
人工智能模型	231	65.63±4.39	66.11±5.51
高年资医师	231	66.04±3.86	66.03±6.33
t值		3.66	0.435
P值		＜0.001	0.664

图3 人工智能模型测量和医师测量髋关节α角和β角的Deming回归分析图。图a为α角的斜率图；图b为α角的回归残差图；图c为β角的斜率图；图d为β角的回归残差图

1	Jackson JC, Runge MM, Nye NS. Common questions about developmental dysplasia of the hip [J]. Am Fam Physician, 2014, 90(12): 843-850.
2	O'Beirne JG, Chlapoutakis K, Alshryda S, et al. International interdisciplinary consensus meeting on the evaluation of developmental dysplasia of the hip [J]. Ultraschall Med, 2019, 40(4): 454-464.
3	王洪杰, 于霞, 高强. 基于深度学习的甲状腺结节自动识别方法在超声图像中的应用 [J]. 中国医疗设备, 2019, 34(10): 72-74, 78.
4	Xie H, Wang N, He M, et al. Using deep learning algorithms to classify fetal brain ultrasound images as normal or abnormal [J]. Ultrasound Obstet Gynecol, 2020, 56(4): 579-587.
5	南嘉格列, 李锐, 王海霞, 等. 基于深度学习的肝包虫病超声图像分型研究 [J]. 深圳大学学报(理工版), 2019, 36(6): 702-708.
6	Buda M, Wildman-Tobriner B, Castor K, et al. Deep learning-based segmentation of nodules in thyroid ultrasound: improving performance by utilizing markers present in the images [J]. Ultrasound Med Biol, 2020, 46(2): 415-421.
7	Al-Dhabyani W, Gomaa M, Khaled H, et al. Dataset of breast ultrasound images [J]. Data Brief, 2020, 28: 104863.
8	余镇, 吴凌云, 倪东, 等. 基于深度学习的胎儿颜面部超声标准切面自动识别 [J]. 中国生物医学工程学报, 2017, 36(3): 267-275.
9	Graf R. Fundamentals of sonographic diagnosis of infant hip dysplasia [J]. J Pediatr Orthop, 1984, 4(6): 735-740.
10	Gyurkovits Z, Sohár G, Baricsa A, et al. Early detection of developmental dysplasia of hip by ultrasound [J]. Hip Int, 2019: 1120700019879687. Online ahead of print.

[1]	张婉微, 秦芸芸, 蔡绮哲, 林明明, 田润雨, 金姗, 吕秀章. 心肌收缩早期延长对非ST段抬高型急性冠脉综合征患者冠状动脉严重狭窄的预测价值[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1016-1022.
[2]	任书堂, 刘晓程, 张亚东, 孙佳英, 陈萍, 周建华, 龙进, 黄云洲. 左心室辅助装置支持下单纯收缩期主动脉瓣反流的超声心动图特征[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1023-1028.
[3]	何金梅, 尹立雪, 谭静, 张文军, 王锐, 任梅, 廖明娇. 超声心肌做功技术对2型糖尿病患者潜在左心室心肌收缩功能损伤的评价[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1029-1035.
[4]	薛艳玲, 马小静, 谢姝瑞, 何俊, 夏娟, 何亚峰. 左心声学造影在急性心肌梗死合并室间隔穿孔中的应用价值[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1036-1039.
[5]	吕琦, 惠品晶, 丁亚芳, 颜燕红. 颈动脉斑块易损性的超声造影评估及与缺血性卒中的相关性研究[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1040-1045.
[6]	魏淑婕, 惠品晶, 丁亚芳, 张白, 颜燕红, 周鹏, 黄亚波. 单侧颈内动脉闭塞患者行颞浅动脉-大脑中动脉搭桥术的脑血流动力学评估[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1046-1055.
[7]	武玺宁, 欧阳云淑, 张一休, 孟华, 徐钟慧, 张培培, 吕珂. 胎儿心脏超声检查在抗SSA/Ro-SSB/La抗体阳性妊娠管理中的应用[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1056-1060.
[8]	杨水华, 何桂丹, 覃桂灿, 梁蒙凤, 罗艳合, 李雪芹, 唐娟松. 胎儿孤立性完全型肺静脉异位引流的超声心动图特征及高分辨率血流联合时间-空间相关成像的应用[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1061-1067.
[9]	张璇, 马宇童, 苗玉倩, 张云, 吴士文, 党晓楚, 陈颖颖, 钟兆明, 王雪娟, 胡淼, 孙岩峰, 马秀珠, 吕发勤, 寇海燕. 超声对Duchenne肌营养不良儿童膈肌功能的评价[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1068-1073.
[10]	张宝富, 俞劲, 叶菁菁, 俞建根, 马晓辉, 刘喜旺. 先天性原发隔异位型肺静脉异位引流的超声心动图诊断[J]. 中华医学超声杂志（电子版）, 2023, 20(10): 1074-1080.
[11]	丁雷, 罗文, 杨晓, 庞丽娜, 张佩蒂, 刘海静, 袁佳妮, 刘瑾. 高帧频超声造影在评价C-TIRADS 4-5类甲状腺结节成像特征中的应用[J]. 中华医学超声杂志（电子版）, 2023, 20(09): 887-894.
[12]	张茜, 陈佳慧, 高雪萌, 赵傲雪, 黄瑛. 基于高帧频超声造影的影像组学特征鉴别诊断甲状腺结节良恶性的价值[J]. 中华医学超声杂志（电子版）, 2023, 20(09): 895-903.
[13]	冯冰, 邹秋果, 梁振波, 卢艳明, 曾奕, 吴淑苗. 老年非特殊型浸润性乳腺癌超声征象与分子生物学指标的临床研究[J]. 中华普外科手术学杂志(电子版), 2024, 18(01): 48-51.
[14]	赵文毅, 邹冰子, 蔡冠晖, 刘永志, 温红. 超声应变力弹性成像联合MRI-DWI靶向引导穿刺在前列腺病变诊断中的应用[J]. 中华临床医师杂志(电子版), 2023, 17(9): 988-994.
[15]	薛念余, 张盛敏, 吴凌恒, 沙蕾, 童揽月, 沈崔琴, 李朝军, 杜联芳. 研究血清胆红素对2型糖尿病患者心脏结构发生改变前心肌功能的影响[J]. 中华临床医师杂志(电子版), 2023, 17(9): 1004-1009.

阅读次数

全文

摘要

选择文件类型/文献管理软件名称

选择包含的内容

Use of deep learning-based artificial intelligence to measure the centred hip joint of infants

模态框（Modal）标题

选择文件类型/文献管理软件名称

选择包含的内容

Use of deep learning-based artificial intelligence to measure the centred hip joint of infants