Please wait a minute...
J Zhejiang Univ (Med Sci)  2020, Vol. 49 Issue (5): 548-555    DOI: 10.3785/j.issn.1008-9292.2020.10.01
    
Artificial intelligence technology in cardiac auscultation screening for congenital heart disease: present and future
XU Weize(),YU Kai,XU Jiajun,YE Jingjing,LI Haomin,SHU Qiang*()
The Heart Center, Children's Hospital, Zhejiang University School of Medicine, National Clinical Research Center for Child Health, National Regional Medical Center for Children, Hangzhou 310052, China
Download: HTML( 23 )   PDF(1329KB)
Export: BibTeX | EndNote (RIS)      

Abstract  

The electronic stethoscope combined with artificial intelligence (AI) technology has realized the digital acquisition of heart sounds and intelligent identification of congenital heart disease, which provides objective basis for heart sound auscultation and improves the accuracy of congenital heart disease diagnosis. At the present stage, the AI based cardiac auscultation technique mainly focuses on the research of AI algorithms, and the researchers have designed and summarized a variety of effective algorithms based on the characteristics of cardiac audio data, among which the mel-frequency cepstral coefficients (MFCC) is the most effective one, and widely used in the cardiac auscultation. However, the current cardiac sound analysis techniques are based on specific data sets, and have not been validated in clinic, so the performance of algorithms need to be further verified. The lack of heart sound data, especially the high-quality, standardized, publicly available heart sound database with disease labeling, further restricts the development of heart sound diagnostic analysis and its application in screening. Therefore, expert consensus is necessary in establishing an authoritative heart sound database and standardizing the heart sound auscultation screening process for congenital heart disease. This paper provides an overview of the research and application status of auscultation algorithm and hardware equipment based on AI in auscultation screening of congenital heart disease, and puts forward the problems to be solved in clinical application of AI auscultation screening technology.



Key wordsHeart defects, congenital      Neonatal screening      Heart auscultation      Artificial intelligence     
Received: 27 August 2020      Published: 19 November 2020
CLC:  R44  
Corresponding Authors: SHU Qiang     E-mail: 120heart@zju.edu.cn;shuqiang@zju.edu.cn
Cite this article:

XU Weize,YU Kai,XU Jiajun,YE Jingjing,LI Haomin,SHU Qiang. Artificial intelligence technology in cardiac auscultation screening for congenital heart disease: present and future. J Zhejiang Univ (Med Sci), 2020, 49(5): 548-555.

URL:

http://www.zjujournals.com/med/10.3785/j.issn.1008-9292.2020.10.01     OR     http://www.zjujournals.com/med/Y2020/V49/I5/548


先天性心脏病心音听诊筛查的人工智能技术应用现状

近年来,电子听诊器结合人工智能技术实现了心音的数字化采集和先天性心脏病的智能识别,为心音听诊提供了客观依据,提高了先天性心脏病诊断的准确率。现阶段基于人工智能技术的智能听诊技术主要侧重于人工智能算法的研究,国内外学者也针对心音音频数据的特点设计总结了多种有效算法,其中梅尔频率倒谱系数(MFCC)是最常用且有效的心音特征,被广泛应用于智能听诊技术中。然而,当前心音智能听诊技术均基于筛选的特定数据集实现,并且尚未在实际临床环境中基于大样本进行实验验证,因此各个算法的实际临床应用表现尚待进一步验证。心音数据匮乏,特别是高质量、标准化、带疾病标注且公开的心音数据库的缺失,进一步制约了心音智能诊断分析技术的发展和听诊筛查的应用。因此,相关医疗单位应当组织有关专家共同建立先天性心脏病心音听诊筛查的专家共识和标准化心音听诊筛查流程,并以此建立权威心音数据库。本文就现阶段基于人工智能的听诊算法和硬件设备在先天性心脏病听诊筛查中的研究及应用进行综述,提出人工智能心音听诊筛查技术在临床应用中有待解决的问题。


关键词: 心脏缺损, 先天性,  新生儿筛查,  心脏听诊,  人工智能 
Fig 1 Diagram of heart sound waveform[4]
Fig 2 The principal processes of artificial intelligence-based heart sound auscultation
产品名称 功能 优点 缺点
Littmann 3200电子听诊器 具备环境消噪技术和蓝牙技术;
具备听诊助手客户端软件;
具备多种听诊模式,能实现全速或半速听诊
心音质量高;
浏览查看方便
无人工智能算法
eMurmur ID智能系统 具备美国食品药品监督管理局认证的人工智能心脏杂音分类算法;
集算法、手机应用程序、门户网站为一体
能够实现心音智能听诊;
浏览操作简单;
支持跨平台操作
无自主研发硬件设备, 依赖第三方电子听诊器
CliniCloud智能听诊器 具备红外线和蓝牙技术;
具备苹果iOS和安卓客户端应用程序;
集成医生点播服务;
具备数字健康工具包,实现健康跟踪和心音回溯
适合家庭使用;
操作简单;
数据管理方便;
有利于医生回溯患者数据
无人工智能算法, 医生远程听诊实时性不确定
The One智能听诊器 超100倍声音放大功能;
具备Thinklink移动套件,实现听诊器与手机、电脑及其他设备连接;
具备移动端应用程序
音质好;
工业设计优;
在医生交流、教育、培训等方面体验友好
无人工智能算法
云听G200智能听诊器 具备蓝牙传输技术;
具备手机、电脑客户端软件;
能够根据需求定制接口;
提供人工智能听诊功能
能够实现智能听诊;
数据传输浏览方便;
售价便宜
软件生态系统欠完善
Tab 1 An introduction to the function, advantages and disadvantages of electronic stethoscope equipment and systems
[1]   黄国英 . 我国开展新生儿先天性心脏病筛查的重要性[J]. 中华儿科杂志, 2017, 55 (4): 241- 243
HUANG Guoying . Importance of neonatal screening for congenital heart disease in China[J]. Chinese Journal of Pediatrics, 2017, 55 (4): 241- 243
doi: 10.3760/cma.j.issn.0578-1310.2017.04.001
[2]   周爱卿 . 先天性心脏病诊断思路和检查方法选择[J]. 诊断学理论与实践, 2006, 5 (3): 276- 278
ZHOU Aiqing . Thinking and selection of methods in diagnosis of congenital heart disease[J]. Journal of Diagnostics:Concepts & Practice, 2006, 5 (3): 276- 278
doi: 10.3969/j.issn.1671-2870.2006.03.029
[3]   郭腾飞 . 心音信号的测量与处理[J]. 南北桥, 2010, (9): 44- 46
GUO Tengfei . Heart sounds signal's measurement and processing[J]. North South Bridge, 2010, (9): 44- 46
doi: 10.3969/j.issn.1672-0407.2010.09.020
[4]   SPRINGER D B , TARASSENKO L , CLIFFORD G D . Logistic Regression-HSMM-based heart sound segmentation[J]. IEEE Trans Biomed Eng, 2016, 63 (4): 822- 832
doi: 10.1109/TBME.2015.2475278
[5]   谢梅兰.心脏杂音分级量化研究及心脏能量分析[D].重庆: 重庆大学, 2010.
XIE Meilan. The study of heart murmur grading quantification and cardiac energy[D]. Chongqing: Chongqing University, 2010. (in Chinese)
[6]   GHOSH S K , PONNALAGU R N , TRIPATHY R K et al. Automated detection of heart valve diseases using chirplet transform and multiclass composite classifier with PCG signals[J]. Comput Biol Med, 2020, 118 103632
doi: 10.1016/j.compbiomed.2020.103632
[7]   DENG M , MENG T , CAO J et al. Heart sound classification based on improved MFCC features and convolutional recurrent neural networks[J]. Neural Netw, 2020, 130 22- 32
doi: 10.1016/j.neunet.2020.06.015
[8]   谭朝文, 王威廉, 宗容 et al. 卷积神经网络应用于先心病心音信号分类研究[J]. 计算机工程与应用, 2019, 55 (12): 174- 180
TAN Zhaowen , WANG Weilian , ZONG Rong et al. Research on classification of congenital heart disease heart sound signal using convolutional neural network[J]. Computer Engineering and Application, 2019, 55 (12): 174- 180
doi: 10.3778/j.issn.1002-8331.1807-0115
[9]   谭朝文, 王威廉, 宗容 et al. 基于卷积神经网络的先心病心音信号分类算法[J]. 生物医学工程学杂志, 2019, 36 (5): 728- 736, 744
TAN Zhaowen , WANG Weilian , ZONG Rong et al. Classification of heart sound signals in congenital heart disease based on convolutional neural network[J]. Journal of Biomedical Engineering, 2019, 36 (5): 728- 736, 744
doi: 10.7507/1001-5515.201806031
[10]   ZHU L L , PAN J H , SHI J H et al. Research on recognition of CHD heart sound using MFCC and LPCC[J]. JPhCS, 2019, 1169 (1): 012011
doi: 10.1088/1742-6596/1169/1/012011
[11]   陈洁, 侯海良, 罗良才 et al. 基于双门限的第一、第二心音自动识别方法[J]. 计算机工程, 2012, 38 (16): 174- 177, 181
CHEN Jie , HOU Hailiang , LUO Liangcai et al. Automatic identification method for the first and second heart sound based on double-threshold[J]. Computer Engineering, 2012, 38 (16): 174- 177, 181
doi: 10.3969/j.issn.1000-3428.2012.16.045
[12]   房玉, 江钟伟, 王海滨 et al. 一种先天性心脏病杂音分割及分析方法[J]. 北京生物医学工程, 2018, 37 (2): 151- 156
FANG Yu , JIANG Zhongwei , WANG Haibin et al. A heart murmur segmentation and analysis method for congenital heart disease[J]. Beijing Biomedical Engineering, 2018, 37 (2): 151- 156
doi: 10.3969/j.issn.1002-3208.2018.02.007
[13]   侯雷静, 郭婷婷, 孙燕 et al. 面向心音分割的个性化高斯混合建模方法[J]. 声学学报, 2019, 44 (1): 22- 29
HOU Leijing , GUO Tingting , SUN Yan et al. A personalized Gaussian mixture model modeling method for heart sound segmentation[J]. Acta Acustica, 2019, 44 (1): 22- 29
doi: 10.15949/j.cnki.0371-0025.2019.01.003
[14]   FERNANDO T , GHAEMMAGHAMI H , DENMAN S et al. Heart sound segmentation using bidirectional LSTMs with attention[J]. IEEE J Biomed Health Inform, 2020, 24 (6): 1601- 1609
doi: 10.1109/JBHI.2019.2949516
[15]   AZIZ S , KHAN M U , ALHAISONI M et al. Phonocardiogram signal processing for automatic diagnosis of congenital heart disorders through fusion of temporal and cepstral features[J]. Sensors (Basel), 2020, 20 (13):
doi: 10.3390/s20133790
[16]   SON G Y , KWON S . Classification of heart sound signal using multiple features[J]. Applied Sciences, 2018, 8 (12): 2344
doi: 10.3390/app8122344
[17]   PATIDAR S , PACHORI R B , GARG N . Automatic diagnosis of septal defects based on tunable-Q wavelet transform of cardiac sound signals[J]. Expert Syst Appl, 2015, 42 (7): 3315- 3326
doi: 10.1016/j.eswa.2014.11.046
[18]   OH S L , JAHMUNAH V , OOI C P et al. Classification of heart sound signals using a novel deep WaveNet model[J]. Comput Methods Programs Biomed, 2020, 196 105604
doi: 10.1016/j.cmpb.2020.105604
[19]   NILANON T, YAO J, HAO J, et al. Normal/abnormal heart sound recordings classification using convolutional neural network[C]. IEEE: 2016 Computing in Cardiology Conference (CinC), 2016: 585-588.
[20]   SUJADEVI V G, SOMAN K P, VINAYAKUMAR R, et al. Deep models for phonocardiography (PCG) classification[C]. IEEE: 2017 International Conference on Intelligent Communication and Computational Techniques (ICCT), 2017: 211-216. DOI: 10.1109/INTELCCT.2017.8324047.
[21]   LIU C , SPRINGER D , LI Q et al. An open access database for the evaluation of heart sound algorithms[J]. Physiol Meas, 2016, 37 (12): 2181- 2213
doi: 10.1088/0967-3334/37/12/2181
[22]   SUN S , JIANG Z , WANG H et al. Automatic moment segmentation and peak detection analysis of heart sound pattern via short-time modified Hilbert transform[J]. Comput Methods Programs Biomed, 2014, 114 (3): 219- 230
doi: 10.1016/j.cmpb.2014.02.004
[23]   SUN S . An innovative intelligent system based on automatic diagnostic feature extraction for diagnosing heart diseases[J]. Knowl-based Syst, 2015, 75 224- 238
doi: 10.1016/j.knosys.2014.12.001
[24]   PATIDAR S , PACHORI R B . Segmentation of cardiac sound signals by removing murmurs using constrained tunable-Q wavelet transform[J]. Biomed Signal Proces, 2013, 8 (6): 559- 567
doi: 10.1016/j.bspc.2013.05.004
[1] HU Zhenzhen, YANG Jianbin, HU Lingwei, ZHAO Yunfei, ZHANG Chao, YANG Rulai, HUANG Xinwen. Screening and clinical analysis of isovaleric acidemia newborn in Zhejiang province[J]. J Zhejiang Univ (Med Sci), 2020, 49(5): 556-564.
[2] HUANG Shumin,ZHAO Zhengyan. Advances in newborn screening and immune system reconstitution of severe combined immunodeficiency[J]. J Zhejiang Univ (Med Sci), 2019, 48(4): 351-357.
[3] WU Dingwen,LU Bin,YANG Jianbin,YANG Rulai,HUANG Xinwen,TONG Fan,ZHENG Jing,ZHAO Zhengyan. Genetic analysis of newborns with abnormal metabolism of 3-hydroxyisovalerylcarnitine[J]. J Zhejiang Univ (Med Sci), 2019, 48(4): 390-396.
[4] TONG Fan,YANG Rulai,LIU Chang,WU Dingwen,ZHANG Ting,HUANG Xinwen,HONG Fang,QIAN Guling,HUANG Xiaolei,ZHOU Xuelian,SHU Qiang,ZHAO Zhengyan. Screening for hereditary tyrosinemia and genotype analysis in newborns[J]. J Zhejiang Univ (Med Sci), 2019, 48(4): 459-464.
[5] XU Jingjing, TAN Yanbin, ZHANG Minming. Medical imaging in tumor precision medicine: opportunities and challenges[J]. J Zhejiang Univ (Med Sci), 2017, 46(5): 455-461.
[6] DONG Fei, LI Qian, JIANG Biao, ZENG Qiang, HUA Jianming, ZHANG Minming. Quantitative analysis of enhanced MRI features for predicting epidermal growth factor receptor gene amplification in glioblastoma multiforme with radiomic method[J]. J Zhejiang Univ (Med Sci), 2017, 46(5): 492-497.
[7] HUANG Xinwen, ZHANG Yu, HONG Fang, ZHENG Jing, YANG Jianbin, TONG Fan, MAO Huaqing, HUANG Xiaolei, ZHOU Xuelian, YANG Rulai, ZHAO Zhengyan. Screening for amino acid metabolic disorders of newborns in Zhejiang province:prevalence, outcome and follow-up[J]. J Zhejiang Univ (Med Sci), 2017, 46(3): 233-239.
[8] ZHU Hui, MIAO Zhengyou, QIAN Yeqing, LI Hongge, JIN Jinglei, HE Jing, DONG Minyue. Detection of cell-free fetal DNA in maternal plasma for noninvasive prenatal screening of fetal chromosomal aneuploidies in women of advanced maternal age[J]. J Zhejiang Univ (Med Sci), 2017, 46(3): 256-261.
[9] HONG Fang, HUANG Xinwen, ZHANG Yu, YANG Jianbin, TONG Fan, MAO Huaqing, HUANG Xiaolei, ZHOU Xuelian, YANG Rulai, ZHAO Zhengyan. Screening for newborn organic aciduria in Zhejiang province:prevalence, outcome and follow-up[J]. J Zhejiang Univ (Med Sci), 2017, 46(3): 240-247.
[10] ZHENG Jing, ZHANG Yu, HONG Fang, YANG Jianbin, TONG Fan, MAO Huaqing, HUANG Xiaolei, ZHOU Xuelian, YANG Rulai, ZHAO Zhengyan, HUANG Xinwen. Screening for fatty acid oxidation disorders of newborns in Zhejiang province:prevalence, outcome and follow-up[J]. J Zhejiang Univ (Med Sci), 2017, 46(3): 248-255.
[11] LI Enshu, YE Xiaoqun, FANG Li, YE Yinghui. Effect of oxygen concentration on outcome of in-vitro fertilization-embryo transfer[J]. J Zhejiang Univ (Med Sci), 2017, 46(3): 290-294.
[12] LU Han-ti, LI Fu-dong, LIN Jun-fen, HE Fan, SHEN Yi. Construction of early warning model of influenza-like illness in Zhejiang Province based on support vector machine[J]. J Zhejiang Univ (Med Sci), 2015, 44(6): 653-658.
[13] XU Wei-hua,CHEN You-ling,YAN Zheng. Development of expert diagnostic system for common respiratory diseases[J]. J Zhejiang Univ (Med Sci), 2014, 43(2): 252-256.
[14] Jiang Guoping, Kang Manli, He Jin. Direct Measurement of the Ratio of Pulmonary Flow to Systemic Blood Flow Among Children by Echocardiography[J]. J Zhejiang Univ (Med Sci), 1999, 28(3): 120-122.