• 《工程索引》(EI)刊源期刊
  • 中文核心期刊(综合性理工农医类)
  • 中国科技论文统计源期刊
  • 中国科学引文数据库来源期刊

留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

基于一维卷积特征与手工特征融合的集成超限学习机心跳分类方法

许越凡 肖文栋 曹征涛

许越凡, 肖文栋, 曹征涛. 基于一维卷积特征与手工特征融合的集成超限学习机心跳分类方法[J]. 工程科学学报, 2021, 43(9): 1224-1232. doi: 10.13374/j.issn2095-9389.2021.01.12.005
引用本文: 许越凡, 肖文栋, 曹征涛. 基于一维卷积特征与手工特征融合的集成超限学习机心跳分类方法[J]. 工程科学学报, 2021, 43(9): 1224-1232. doi: 10.13374/j.issn2095-9389.2021.01.12.005
XU Yue-fan, XIAO Wen-dong, CAO Zheng-tao. Ensemble extreme learning machine approach for heartbeat classification by fusing 1d convolutional and handcrafted features[J]. Chinese Journal of Engineering, 2021, 43(9): 1224-1232. doi: 10.13374/j.issn2095-9389.2021.01.12.005
Citation: XU Yue-fan, XIAO Wen-dong, CAO Zheng-tao. Ensemble extreme learning machine approach for heartbeat classification by fusing 1d convolutional and handcrafted features[J]. Chinese Journal of Engineering, 2021, 43(9): 1224-1232. doi: 10.13374/j.issn2095-9389.2021.01.12.005

基于一维卷积特征与手工特征融合的集成超限学习机心跳分类方法

doi: 10.13374/j.issn2095-9389.2021.01.12.005
基金项目: 国家重点研发计划课题资助项目(2017YFB1401203);佛山市科技创新专项资金资助项目(BK20AF005)
详细信息
    通讯作者:

    E-mail: czhengtao@126.com

  • 中图分类号: TP182

Ensemble extreme learning machine approach for heartbeat classification by fusing 1d convolutional and handcrafted features

More Information
  • 摘要: 融合手工特征和深度特征,提出了一种集成超限学习机心跳分类方法。手工提取的特征明确地表征了心电信号的特定特性,如相邻心跳时间间隔反映了心跳信号的时域特性,小波系数反映了心跳信号的时频特性。同时设计了一维卷积神经网络对心跳信号特征进行自动提取。基于超限学习机(Extreme leaning machine,ELM),将上述特征融合进行心跳分类。由于ELM初始参数的随机给定可能导致其性能不稳定,进一步提出了一种基于袋装(Bagging)策略的多个ELM集成方法,使分类结果更加稳定且模型泛化能力更强。利用麻省理工心律失常公开数据集对所提方法进行了验证,分类准确率达到了99.02%,实验结果也表明基于融合特征的分类准确率高于基于单独特征的分类准确率。

     

  • 图  1  ELM的基本结构

    Figure  1.  Basic structure of an extreme learning machine (ELM)

    图  2  心跳分类算法总体结构

    Figure  2.  Overall structure of the heartbeat classification algorithm

    图  3  本文提出的1D CNN结构

    Figure  3.  Structure of the proposed 1D convolutional neural network

    图  4  ELM心跳分类结构

    Figure  4.  ELM heartbeat classification structure

    表  1  MIT−BIH数据集的详细描述和划分

    Table  1.   Detailed description and division of the MIT−BIH dataset

    Heartbeat type (abbreviation)AnnotationTotal
    number of
    samples
    Number of training samplesNumber of testing samples
    Normal beat (NOR)N75023975365270
    Left bundle branch block (LBBB)L807232294843
    Right bundle branch block (RBBB)R725529024353
    Atrial premature contraction (APC)A254610181528
    Premature ventricular contraction (PVC)V712928524277
    Paced beat (PACE)/702628104216
    Aberrated atrial premature beat (AP)a1507575
    Ventricular flutter
    wave (VF)
    !472236236
    Fusion of ventricular and normal beat (VFN)F802401401
    Blocked atrial premature beat (BAP)x1939697
    Nodal (junctional) escape beat (NE)j229114115
    Fusion of paced and normal beat (FPN)f982491491
    Ventricular escape
    beat (VE)
    E1065353
    Nodal (junctional) premature beat (NP)J834241
    Atrial escape beat (AE)e1688
    Unclassifiable beat (UN)Q331617
    Total161101172409686021
    下载: 导出CSV

    表  2  每类心跳的召回率和精确率

    Table  2.   Recall and precision for each heartbeat class

    Heartbeat typeNumber of test samplesRecall/%Precision/%
    N6527099.5899.43
    L484399.7199.67
    R435399.6199.34
    A152886.4594.90
    V427797.8095.57
    /421699.6999.41
    a7565.3392.45
    !23693.6492.47
    F40179.5587.40
    x9786.6094.38
    j11587.8371.13
    f49191.6596.98
    E5394.3498.04
    J4190.2497.37
    e812.50100.00
    Q175.8850.00
    Total8602199.0299.02
    下载: 导出CSV

    表  3  混淆矩阵

    Table  3.   Confusion matrix

    Predict labels
    NLRAV/a!FxjfEJeQ
    True labelsN6499693581261082843240001
    L6482900700000001000
    R8043366201000000000
    A1691201321900100610000
    V624124183015180010000
    /5000142030000160000
    a112047049200000000
    !800050122101000000
    F54010270003190000000
    x40014012084100000
    j1201000000010100100
    f16000124000004500000
    E10002000000050000
    J10200000001003700
    e7000000000000010
    Q10010300000020001
    下载: 导出CSV

    表  4  提出的方法与其他方法的比较结果

    Table  4.   Comparison results of the proposed approach with other approaches

    ReferenceFeatures + ClassifierAccuracy/
    %
    Manual features onlyDWT, RR + ELM (Single)98.28
    Deep feature only1D CNN98.50
    Feature fusion
    (Without ensemble)
    DWT, RR, 1D Convolution +
    ELM (Single)
    98.81
    Ye[9]ICA, Wavelet, RR + SVM
    (One-against-one)
    98.72
    Our proposed approachDWT, RR, 1D Convolution + ELM
    (Bagging ensemble)
    99.02
    下载: 导出CSV
  • [1] Thaler M S. The Only EKG Book You’ll Ever Need. 9th Ed. Philadelphia, PA: Lippincott Williams & Wilkins, 2017[2] Gawlowska J, Wranicz J K, Norman J. “Jeff ” Holter (1914—1983). Cardiol J, 2009, 16(4): 386
    [2] Chazal P d, Reilly R B. A patient-adapting heartbeat classifier using ECG morphology and heartbeat interval features. IEEE Trans Biomed Eng, 2006, 53(12): 2535 doi: 10.1109/TBME.2006.883802
    [3] Mondéjar-Guerra V, Novo J, Rouco J, et al. Heartbeat classification fusing temporal and morphological information of ECGs via ensemble of classifiers. Biomed Signal Process Control, 2019, 47: 41 doi: 10.1016/j.bspc.2018.08.007
    [4] Chazal P d, O'Dwyer M, Reilly R B. Automatic classification of heartbeats using ECG morphology and heartbeat interval features. IEEE Trans Biomed Eng, 2004, 51(7): 1196 doi: 10.1109/TBME.2004.827359
    [5] Tuncer T, Dogan S, Pławiak P, et al. Automated arrhythmia detection using novel hexadecimal local pattern and multilevel wavelet transform with ECG signals. Knowl Based Syst, 2019, 186: 104923 doi: 10.1016/j.knosys.2019.104923
    [6] Sahoo S, Kanungo B, Behera S, et al. Multiresolution wavelet transform based feature extraction and ECG classification to detect cardiac abnormalities. Measurement, 2017, 108: 55 doi: 10.1016/j.measurement.2017.05.022
    [7] Afkhami R G, Azarnia G, Tinati M A. Cardiac arrhythmia classification using statistical and mixture modeling features of ECG signals. Pattern Recognit Lett, 2016, 70: 45 doi: 10.1016/j.patrec.2015.11.018
    [8] Marinho L B, Nascimento N M M, Souza J W M, et al. A novel electrocardiogram feature extraction approach for cardiac arrhythmia classification. Future Gener Comput Syst, 2019, 97: 564 doi: 10.1016/j.future.2019.03.025
    [9] Ye C, Vijaya Kumar B V K, Coimbra M T. Heartbeat classification using morphological and dynamic features of ECG signals. IEEE Trans Biomed Eng, 2012, 59(10): 2930 doi: 10.1109/TBME.2012.2213253
    [10] Castillo O, Melin P, Ramírez E, et al. Hybrid intelligent system for cardiac arrhythmia classification with Fuzzy K-Nearest Neighbors and neural networks combined with a fuzzy system. Expert Syst Appl, 2012, 39(3): 2947 doi: 10.1016/j.eswa.2011.08.156
    [11] Girshick R. Fast R−CNN // 2015 IEEE International Conference on Computer Vision (ICCV). Santiago, 2015: 1440
    [12] Huang G, Liu Z, Van Der Maaten L, et al. Densely connected convolutional networks // 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, 2017: 2261
    [13] He K M, Zhang X Y, Ren S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Trans Pattern Anal Mach Intell, 2015, 37(9): 1904 doi: 10.1109/TPAMI.2015.2389824
    [14] Szegedy C, Liu W, Jia Y Q, et al. Going deeper with convolutions // 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, 2015: 1
    [15] Schroff F, Kalenichenko D, Philbin J. FaceNet: A unified embedding for face recognition and clustering // 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, 2015: 815
    [16] Ding C X, Tao D C. Trunk-branch ensemble convolutional neural networks for video-based face recognition. IEEE Trans Pattern Anal Mach Intell, 2018, 40(4): 1002 doi: 10.1109/TPAMI.2017.2700390
    [17] He R, Wu X, Sun Z N, et al. Wasserstein CNN: Learning invariant features for NIR-VIS face recognition. IEEE Trans Pattern Anal Mach Intell, 2019, 41(7): 1761 doi: 10.1109/TPAMI.2018.2842770
    [18] Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation. IEEE Trans Pattern Anal Mach Intell, 2017, 39(4): 640 doi: 10.1109/TPAMI.2016.2572683
    [19] LeCun Y, Bottou L, Bengio Y, et al. Gradient-based learning applied to document recognition. Proc IEEE, 1998, 86(11): 2278 doi: 10.1109/5.726791
    [20] Huang G B, Zhu Q Y, Siew C K. Extreme learning machine: Theory and applications. Neurocomputing, 2006, 70(1-3): 489 doi: 10.1016/j.neucom.2005.12.126
    [21] Huang G B, Chen L, Siew C K. Universal approximation using incremental constructive feedforward networks with random hidden nodes. IEEE Trans Neural Netw, 2006, 17(4): 879 doi: 10.1109/TNN.2006.875977
    [22] Anthimopoulos M, Christodoulidis S, Ebner L, et al. Lung pattern classification for interstitial lung diseases using a deep convolutional neural network. IEEE Trans Med Imaging, 2016, 35(5): 1207 doi: 10.1109/TMI.2016.2535865
    [23] Li J Y, Zhao Y K, Xue Z E, et al. A survey of model compression for deep neural networks. Chin J Eng, 2019, 41(10): 1229

    李江昀, 赵义凯, 薛卓尔, 等. 深度神经网络模型压缩综述. 工程科学学报, 2019, 41(10):1229
    [24] Ji S W, Xu W, Yang M, et al. 3D convolutional neural networks for human action recognition. IEEE Trans Pattern Anal Mach Intell, 2013, 35(1): 221 doi: 10.1109/TPAMI.2012.59
    [25] Bartlett P L. The sample complexity of pattern classification with neural networks: The size of the weights is more important than the size of the network. IEEE Trans Inf Theory, 1998, 44(2): 525 doi: 10.1109/18.661502
    [26] PhysioNet. MIT−BIH Arrhythmias Database [J/OL]. PhysioNet Online (2020-04-15) [2021-01-12] https://physionet.org/files/mitdb/1.0.0
    [27] Moody G B, Mark R G. The impact of the mit-bih arrhythmia database. IEEE Eng Med Biol Mag, 2001, 20(3): 45 doi: 10.1109/51.932724
  • 加载中
图(4) / 表(4)
计量
  • 文章访问数:  178
  • HTML全文浏览量:  90
  • PDF下载量:  35
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-01-12
  • 网络出版日期:  2021-03-10
  • 刊出日期:  2021-09-18

目录

    /

    返回文章
    返回