[1]林 楠,唐凯鹏,牛勇鹏,等.基于双阶段特征提取网络的 ECG 降噪分类算法[J].郑州大学学报(工学版),2024,45(05):61-68.[doi:10.13705/j.issn.1671-6833.2024.05.005]
　LIN Nan,TANG Kaipeng,NIU Yongpeng,et al.An ECG Denoising and Classification Algorithm Based on Two-stage Feature Extraction Network[J].Journal of Zhengzhou University (Engineering Science),2024,45(05):61-68.[doi:10.13705/j.issn.1671-6833.2024.05.005]
点击复制

基于双阶段特征提取网络的 ECG 降噪分类算法()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
45
期数:
2024年05期
页码:
61-68
栏目:
出版日期:
2024-08-08

文章信息/Info

Title:
An ECG Denoising and Classification Algorithm Based on Two-stage Feature Extraction Network
文章编号:
1671-6833(2024)05-0061-08
作者:
林 楠;  唐凯鹏;  牛勇鹏;  谢李鹏
郑州大学 网络空间安全学院,河南 郑州 450003
Author(s):
LIN Nan;  TANG Kaipeng;  NIU Yongpeng;  XIE Lipeng
School of Cyber Science and Engineering, Zhengzhou University, Zhengzhou 450003, China
关键词:
心电信号分类;  心电信号去噪;  残差收缩网络;  软阈值化;  注意力机制
Keywords:
ECG classification;  ECG denoising;  residual shrinkable network;  soft thresholding;  attention mechanism
分类号:
TP183TN911. 7R541. 7
DOI:
10.13705/j.issn.1671-6833.2024.05.005
文献标志码:
A
摘要:
临床采集到的标准 12 导联心电图常含有噪声,影响了心电信号分类结果的准确度,为此提出了一种基于双阶段特征提取网络的心电图(ECG)降噪分类算法。 首先,在空间特征提取阶段,由深度耦合软阈值化去噪方法的残差收缩网络从输入的 12 导联标准心电信号中提取空间特征;其次,在时间特征提取阶段,由长短期记忆网络与注意力机制结合继续从心电信号中提取时间特征;最后,通过全连接网络层融合提取到的空间特征与时间特征,输出 9 个类别的概率预测分布。 在 CPSC2018 数据集上与其他同类型先进分类算法进行了对比实验,验证所提算法的效果,实验结果表明:提出的分类算法在对 9 类 ECG 信号进行分类时平均 F1 分数达到 0. 854,在各项指标上表现更优。 此外,实验证明所提算法在含噪数据中的表现也优于其他主流网络,充分证明了所提算法对于含噪心电信号的降噪分类性能,该算法也可应用于其他类似含噪声生理信号的分析和处理。
Abstract:
Since clinically acquired standard 12-lead ECGs often contain noise, which could affects the accuracy ofthe ECG signal classification results, a noise reduction classification algorithm for ECGs based on a two-stagefeature extraction network was proposed. Firsty, in the spatial feature extraction stage, spatial features wereextracted from the input 12-lead standard ECG signal by a residual contraction network with a deeply coupled softthresholding denoising method. Secondly, in the temporal feature extraction stage, temporal features were extractedfrom the ECG signal by a combination of a long and short-term memory network and an attentional mechanism. Andultimately, the extracted spatial and temporal features were fused through the fully-connected network layer tooutput the probabilistic predictive distributions for the nine categories. In order to verify the effect of the proposedalgorithm, comparison experiments were conducted with other state-of-the-art classification algorithms of the sametype on the CPSC2018 dataset, and the experimental results showed that the proposed classification algorithm couldachieve an average F1 score of 0. 848 when classifying the nine categories of ECG signals, which was a much betterperformance in terms of various indicators. In addition, the experiment proved that the proposed algorithm alsocould outperform other mainstream networks in noise-containing data, which fully demonstrated the noise reductionclassification performance of the proposed algorithm for noise-containing ECG signals. And the algorithm can alsobe applied to other similar noise-containing physiological signals for analysis and processing.

参考文献/References:

[1] ROTH G A, ABATE D, ABATE K H. Global, regional,and national age-sex-specific mortality for 282 causes ofdeath in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study2017[ J] . Lancet, 2018, 392(10159) : 1736-1788.

[2] ROBERT S K, COLLEEN E C. Basis and treatment ofcardiac arrhythmias[M] . Berlin:Springer,2006.
[3] MIEGHEM C, SABBE M, KNOCKAERT D. The clinicalvalue of the ECG in noncardiac conditions [ J ] . Chest,2004, 125(4) : 1561-1576.
[4] STAMKOPOULOS T, DIAMANTARAS K, MAGLAVERASN, et al. ECG analysis using nonlinear PCA neural networksfor ischemia detection [ J ]. IEEE Transactions on SignalProcessing, 1998, 46(11) : 3058-3067.
[5] 朱凌云. 移动心电监护系统 ECG 信号的智能检测与分析方法研究[D] . 重庆: 重庆大学, 2003.
ZHU L Y. The algorithms research of ECG signals intelligent detection & analysis for mobile cardiac telemonitoringsystem[D] . Chongqing: Chongqing University, 2003.
[6] PERLMAN O, KATZ A, AMIT G, et al. Supraventricular tachycardia classification in the 12-lead ECG using atrial waves detection and a clinically based tree scheme[ J] . IEEE Journal of Biomedical and Health Informatics,2016, 20(6) : 1513-1520.
[7] 李润川, 张行进, 陈刚, 等. 基于多特征融合的心搏类型识别研究[ J] . 郑州大学学报(工学版) , 2021, 42(4) : 7-12.
LI R C, ZHANG X J, CHEN G, et al. Research onheartbeat type recognition based on multi-feature fusion[ J] . Journal of Zhengzhou University ( Engineering Science) , 2021, 42(4) : 7-12.
[8] SCHLÄPFER J, WELLENS H J. Computer-interpreted electrocardiograms: benefits and limitations[ J] . Journal ofthe American College of Cardiology, 2017, 70 ( 9 ) :1183-1192.
[9] HANNUN A Y, RAJPURKAR P, HAGHPANAHI M, etal. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neuralnetwork[ J] . Nature Medicine, 2019, 25: 65-69.
[10] RIBEIRO A H, RIBEIRO M H, PAIXÃO G M M, et al.Automatic diagnosis of the 12-lead ECG using a deep neural network [ J ] . Nature Communications, 2020,11: 1760.
[11] 王文刀, 王润泽, 魏鑫磊, 等. 基于堆叠式双向 LSTM的心电图 自 动 识 别 算 法 [ J] . 计 算 机 科 学, 2020, 47(7) : 118-124.
WANG W D, WANG R Z, WEI X L, et al. Automaticrecognition of ECG based on stacked bidirectional LSTM[ J] . Computer Science, 2020, 47(7) : 118-124.
[12] FUKUSHIMA K. Neocognitron: a self-organizing neuralnetwork model for a mechanism of pattern recognition unaffected by shift in position [ J] . Biological Cybernetics,1980, 36(4) : 193-202.
[13] GOLLER C, KUCHLER A. Learning task-dependent distributed representations by backpropagation through structure [ C ] ∥ Proceedings of International Conference onNeural Networks. Piscataway: IEEE, 1996: 347-352.
[14] LIU F, ZHOU X S, CAO J L, et al. A LSTM and CNNbased assemble neural network framework for arrhythmiasclassification[C]∥ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2019: 1303-1307.
[15] SHAKER A M, TANTAWI M, SHEDEED H A, et al.Generalization of convolutional neural networks for ECGclassification using generative adversarial networks [ J ] .IEEE Access, 2020, 8: 35592-35605.
[16] YAO Q H, FAN X M, CAI Y P, et al. Time-incrementalconvolutional neural network for arrhythmia detection invaried-length electrocardiogram[C]∥2018 IEEE 16th IntlConf on Dependable, Autonomic and Secure Computing,16th Intl Conf on Pervasive Intelligence and Computing,4th Intl Conf on Big Data Intelligence and Computing andCyber Science and Technology Congress. Piscataway:IEEE, 2018: 754-761.
[17] CHEN C Y, LIN Y T, LEE S J, et al. Automated ECGclassification based on 1D deep learning network [ J ] .Methods, 2022, 202: 127-135.
[18] YE X H, HUANG Y Q, LU Q. Automatic multichannelelectrocardiogram record classification using XGBoost fusion model[J]. Frontiers in Physiology,2022,13: 840011.
[19] ZHAO M H, ZHONG S S, FU X Y, et al. Deep residualshrinkage networks for fault diagnosis[J]. IEEE Transactionson Industrial Informatics, 2020, 16(7): 4681-4690.
[20] DONOHO D L. De-noising by soft-thresholding[ J] . IEEETransactions on Information Theory, 1995, 41(3) : 613-627.
[21] The China physiological signal challenge 2018. [ EB /OL] . (2018 - 10 - 17 ) [ 2024 - 02 - 10 ] . https:∥2018.icbeb. org / Challenge. html.
[22] SIMONYAN K, ZISSERMAN A. Very deep convolutionalnetworks for large-scale image recognition [ EB / OL ] .(2014 - 09 - 04 ) [ 2024 - 02 - 10 ] . https: ∥ doi. org /10. 48550 / arXiv. 1409. 1556.
[23] HE K M, ZHANG X Y, REN S Q, et al. Deep residuallearning for image recognition [ C] ∥2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778.
[24] DONG Y F, CAI W Q, QIU L S, et al. Detection of arrhythmia in 12-lead varied-length ECG using multi-branchsignal fusion network [ J ] . Physiological Measurement,2022, 43(10) : 105009.

更新日期/Last Update: 2024-09-02