基于深度学习的超声影像诊断对终末期慢性肾病的价值

doi:10.3877/cma.j.issn.1672-6448.2021.06.012

中华医学超声杂志（电子版） ›› 2021, Vol. 18 ›› Issue (06) : 611 -615. doi: 10.3877/cma.j.issn.1672-6448.2021.06.012

泌尿生殖系统超声影像学

基于深度学习的超声影像诊断对终末期慢性肾病的价值

李广涵¹, 刘健¹, 马立勇², 董梦超², 武敬平¹, 张波¹, 李文歌³, 郑敏¹^,^†(

)

^1. 100029　北京，中日友好医院超声医学科
^2. 100029　北京，中日友好医院肾病科
^3. 264209　哈尔滨工业大学（威海）信息科学与工程学院

收稿日期:2021-03-15 出版日期:2021-06-01
通信作者: 郑敏
基金资助:
国家重点研发计划政府间国际科技创新合作重点专项(2017YFE0110500); 山东省自然科学基金项目(ZR2018MF026); 山东省重点研发计划(2019GGX101054)

Deep learning-based model for ultrasound diagnosis of end-stage chronic kidney disease

Guanghan Li¹, Jian Liu¹, Liyong Ma², Mengchao Dong², Jingping Wu¹, Bo Zhang¹, Wenge Li³, Min Zheng¹()

^1. Department of Ultrasound Medicine, China-Japan Friendship Hospital, Beijing 100029, China
^2. School of Information Science and Engineering, Harbin Institute of Technology, Weihai 264209, China
^3. Department of Nephrology, China-Japan Friendship Hospital, Beijing 100029, China

Received:2021-03-15 Published:2021-06-01
Corresponding author: Min Zheng

全文 ( ) 下载PDF ( ) 导出引用

引用本文：

李广涵, 刘健, 马立勇, 董梦超, 武敬平, 张波, 李文歌, 郑敏. 基于深度学习的超声影像诊断对终末期慢性肾病的价值[J]. 中华医学超声杂志（电子版）, 2021, 18(06): 611-615.

Guanghan Li, Jian Liu, Liyong Ma, Mengchao Dong, Jingping Wu, Bo Zhang, Wenge Li, Min Zheng. Deep learning-based model for ultrasound diagnosis of end-stage chronic kidney disease[J]. Chinese Journal of Medical Ultrasound (Electronic Edition), 2021, 18(06): 611-615.

目的

探讨基于深度学习的卷积神经网络模型DenseNet121对终末期肾病的诊断价值。

方法

回顾性选择2019年1月1日至9月30日期间中日友好医院诊断为终末期肾病的489张肾超声影像和对照组450张健康肾超声影像。采用深度卷积神经网络模型DenseNet121进行网络的训练和验证，把是否为终末期肾病作为参考标准。然后将深度学习模型的诊断结果与专业影像医师的诊断结果进行比较。以受试者工作特征曲线（ROC）评价深度学习模型的诊断性能，以准确性、特异度、敏感度和曲线下面积（AUC）作为衡量指标比较深度学习模型和专业影像医师的诊断性能，采用Delong方法比较2种诊断方式AUC的差异。

结果

专业医师对终末期肾病诊断的准确性为89.36%，敏感度为81.63%，特异度为97.77%，AUC为0.897。基于深度学习的卷积神经网络模型对终末期肾病诊断的准确性为93.51%，敏感度为96.12%，特异度为90.66%，AUC为0.934；与专业医师相比，具有更高的诊断能力（Z=3.034，P=0.002）。

结论

基于深度学习的超声诊断方法显示了较高的诊断性能，有潜力辅助专业影像医师，进行终末期肾病的诊断。

关键词: 慢性肾病, 超声, 深度学习, 诊断

Objective

To evaluate the diagnostic value of the convolution neural network model DenseNet121 based on deep learning in the diagnosis of end-stage renal disease (ESRD).

Methods

In this retrospective study, 489 kidney ultrasound images of patients diagnosed with end-stage renal disease from January 1, 2019 to September 30, 2019 and 450 kidney ultrasound images of healthy controls were selected at China-Japan Friendship Hospital. The deep learning-based supervised convolutional neural network model DenseNet121 was used for network training and verification. According to whether it was end-stage renal disease or not, the prediction results of the deep learning-based model were compared with the prediction results of professional imaging physicians. Receiver operating characteristic (ROC) curve analysis was used to evaluate the performance of the deep learning-based model, the accuracy, specificity, sensitivity, and area under the curve (AUC) were used as metrics to compare the performance of the deep learning-based model and professional imaging physicians, and Delong was used to compare the difference of AUC.

Results

The prediction accuracy of professional imaging physicians for end-stage renal disease was 89.36%, the sensitivity was 81.63%, the specificity was 97.77%, and the AUC was 0.897. The prediction accuracy of the deep learning-based convolution neural network model DenseNet121 for end-stage renal disease was 93.51%, the sensitivity was 96.12%, the specificity was 90.66%, and the AUC was 0.934. Compared with professional physicians, the DenseNet121 model had higher diagnostic ability (Z=3.034, P=0.002).

Conclusion

The ultrasonic diagnosis method based on deep learning shows high diagnostic performance, and it has the potential to assist professional imaging physicians in the diagnosis of end-stage renal disease.

Key words: Chronic kidney disease, Ultrasound, Deep learning, Diagnosis

图1 肾超声图像的感兴趣区域标注结果。图a为控制组，图b为对照组

图2 卷积神经网络DenseNet121网络架构

图3 基于深度学习模型和专业医师人工读图对终末期肾病超声诊断的受试者操作特征曲线

表1 深度学习模型和专业医师人工读图的5次诊断效能表（张）

轮次	深度学习模型				人工读图
轮次	TP	FN	FP	TN	TP	FN	FP	TN
1	92	6	5	85	80	18	2	88
2	94	4	6	84	82	16	3	87
3	98	0	14	76	84	14	1	89
4	94	4	7	83	79	19	2	88
5	93	5	10	80	75	23	2	88

表1 深度学习模型和专业医师人工读图的5次诊断效能表（张）

轮次	深度学习模型				人工读图
轮次	TP	FN	FP	TN	TP	FN	FP	TN
1	92	6	5	85	80	18	2	88
2	94	4	6	84	82	16	3	87
3	98	0	14	76	84	14	1	89
4	94	4	7	83	79	19	2	88
5	93	5	10	80	75	23	2	88

1	Alnazer I, Bourdon P, Urruty T, et al. Recent advances in medical image processing for the evaluation of chronic kidney disease [J]. Med Image Anal, 2021, 69: 101960.
2	胡盛寿, 高润霖, 刘力生, 等. 《中国心血管病报告2018》概要 [J]. 中国循环杂志, 2019, 34(3): 209-220.
3	Petrucci I, Clementi A, Sessa C, et al. Ultrasound and color Doppler applications in chronic kidney disease [J]. J Nephrol, 2018, 31(6): 863-879.
4	卓莉, 邹古明, 李文歌. 人工智能在肾脏病理诊断中的应用 [J/OL]. 中华肾病研究电子杂志, 2020, 9(3): 135-137.
5	解添淞, 周正荣. 人工智能及影像组学在腹部肿瘤中的应用进展 [J]. 中华放射学杂志, 2020, 54(4): 376-379.
6	刘帮燕, 赵丽霞, 郑曙光, 等. 超声多参数评分诊断慢性肾病 [J]. 中国医学影像技术, 2021, 37(2): 273-277.
7	LeCun Y, Bengio Y, Hinton G. Deep learning [J]. Nature, 2015, 521(7553): 436-444.
8	Xie G, Chen T, Li Y, et al. Artificial intelligence in nephrology: how can artificial intelligence augment nephrologists' intelligence? [J]. Kidney Dis (Basel), 2020, 6(1): 1-6.
9	Ma L, Ma C, Liu Y, et al. Thyroid diagnosis from SPECT images using convolutional neural network with optimization [J]. Comput Intell Neurosci, 2019, 2019: 6212759.
10	Brattain LJ, Telfer BA, Dhyani M, et al. Machine learning for medical ultrasound: status, methods, and future opportunities [J]. Abdom Radiol (NY), 2018, 43(4): 786-799.
11	Yi J, Kang H, Kwon J, et al. Technology trends and applications of deep learning in ultrasonography: image quality enhancement, diagnostic support, and improving workflow efficiency [J]. Ultrasonography, 2021, 40(1): 7-22.
12	Kuo C, Chang C, Liu K, et al. Automation of the kidney function prediction and classification through ultrasound-based kidney imaging using deep learning [J]. NPJ Digital Med, 2019, 2: 29.
13	Zheng Q, Furth S, Tasian G, et al. Computer-aided diagnosis of congenital abnormalities of the kidney and urinary tract in children based on ultrasound imaging data by integrating texture image features and deep transfer learning image features [J]. J Pediatr Urolo, 2019, 15(1): 75.e1-75.e7.
14	Ma F, Sun T, Liu L, et al. Detection and diagnosis of chronic kidney disease using deep learning-based heterogeneous modified artificial neural network [J]. Future Generation Computer Systems, 2020, 111: 17-26.
15	Wu Y, Yi Z. Automated detection of kidney abnormalities using multi-feature fusion convolutional neural networks [J]. Knowledge-Based Systems, 2020, 200(2): 105873.
16	Chen C, Pai T, Hsu H, et al. Prediction of chronic kidney disease stages by renal ultrasound imaging [J]. Enterprise Information Systems, 2019, 14(2): 178-195.
17	Sudharson S, Kokil P. An ensemble of deep neural networks for kidney ultrasound image classification [J]. Comput Methods Programs Biomed, 2020, 197: 105709.
18	李广涵, 刘建, 武敬平, 等. 基于支持向量机多模态超声模型诊断肾疾病 [J]. 中国医学影像技术, 2020, 36(6): 898-902.
19	江卫星, 郑闪, 寿建忠, 等. 人工智能在肾细胞癌诊断中的研究现状 [J]. 中华泌尿外科杂志, 2020, 41(3): 233-236.

[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]	彭旭, 邵永孚, 李铎, 邹瑞, 邢贞明. 结肠肝曲癌的诊断和外科治疗[J]. 中华普外科手术学杂志(电子版), 2024, 18(01): 108-110.
[12]	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.
[13]	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.
[14]	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.
[15]	袁媛, 赵良平, 刘智慧, 张丽萍, 谭丽梅, 閤梦琴. 子宫内膜癌组织中miR-25-3p、PTEN的表达及与病理参数的关系[J]. 中华临床医师杂志(电子版), 2023, 17(9): 1016-1020.

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

Deep learning-based model for ultrasound diagnosis of end-stage chronic kidney disease

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

Deep learning-based model for ultrasound diagnosis of end-stage chronic kidney disease