结合CNN与双向LSTM的心律失常分类

doi:10.3778/j.issn.1673-9418.2010053

摘要/Abstract

摘要：

心律失常是心血管疾病中常见的病症之一，实现心律失常的自动分类对心血管疾病的诊治具有重要意义。基于一维心电信号的心律失常分类方法以若干节拍作为输入，通过模型提取特征并用于分类。针对现有方法预处理时间成本高以及未按医疗仪器促进协会（AAMI）标准分类等问题，提出了一种基于原始一维心电信号并按照AAMI推荐标准类别进行心律失常自动分类的方法。该方法首先利用卷积神经网络（CNN）学习心电信号的形态特征，之后通过双向长短期记忆网络（BLSTM）获取特征中的上下文依赖关系，最后借助softmax函数完成分类任务。方法采用mish函数作为激活函数，使得模型在训练中更为稳定。在公开数据库MIT-BIH上进行五折交叉验证，评估结果达到了99.11%的平均准确率，表明该模型可以有效地提取心电信号的特征，适用于监测系统中心律失常疾病的诊断。

关键词: 心电信号, 心律失常, 卷积神经网络（CNN）, 双向长短期记忆网络（BLSTM）

Abstract:

Arrhythmia is one of the common diseases in cardiovascular diseases. Automatic classification of arrhythmia is of great significance to the diagnosis and treatment of cardiovascular diseases. The arrhythmia classification method based on one-dimensional ECG signals takes several beats as input and extracts features from the model for classification. Aiming at the problems of high preprocessing cost and failure to classify according to AAMI (Association for the Advancement of Medical Instrumentation) recommended standards, a method for automatic classification of arrhythmias based on original one-dimensional ECG signals and according to AAMI recommended standards is proposed. This method first uses the convolutional neural network (CNN) to learn the morphological characteristics of the ECG signals, then obtains the context dependence of the characteristics through the bidirectional long short-term memory (BLSTM) network, and finally uses the softmax function to complete the classification task. The mish function is used as the activation function to make the model more stable in training. The five-folds cross-validation is carried out on MIT-BIH, and the evaluation results achieve an average accuracy of 99.11%, indicating that the model can effectively extract the characteristics of the ECG signal and is suitable for the diagnosis of arrhythmia in the monitoring system.

Key words: ECG, arrhythmia, convolutional neural network (CNN), bidirectional long short-term memory (BLSTM) network

李兴秀, 唐建军, 华晶. 结合CNN与双向LSTM的心律失常分类[J]. 计算机科学与探索, 2021, 15(12): 2353-2361.

LI Xingxiu, TANG Jianjun, HUA Jing. Arrhythmia Classification Based on CNN and Bidirectional LSTM[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(12): 2353-2361.

参考文献

[1] GOLRIZKHATAMI Z, ACAN A. ECG classification using three-level fusion of different feature descriptors[J]. Expert Systems with Applications, 2018, 114(12): 54-64.
[2] YE C, VIJAYA KUMAR B V K, COIMBRA M T. Heartbeat classification using morphological and dynamic features of ECG signals[J]. IEEE Transactions on Biomedical Engineering, 2012, 59(10): 2930-2941.
[3] CHEN C, HUA Z C, ZHANG R Q, et al. Automated arrhythmia classification based on a combination network of CNN and LSTM[J]. Biomedical Signal Processing and Control, 2020, 57: 101819.
[4] JANGRA M, DHULL S K, SINGH K K. ECG beat classifiers: a journey from ANN to DNN[J]. Procedia Computer Science, 2020, 167: 747-759.
[5] SONG X, YANG G, WANG K, et al. Short term ECG classification with residual-concatenate network and metric learning[J]. Multimedia Tools and Applications, 2020, 79(31): 22325-22336.
[6] Testing and reporting performance results of cardiac rhythm and ST segment measurement algorithms: ANSI/EC57-2012[S]. American National Standards Institute, 2012.
[7] MARTIS R, ACHARYA U R, LIM C. ECG beat classification using PCA, LDA, ICA and discrete wavelet transform[J]. Biomedical Signal Processing and Control, 2013, 8(5): 437-448.
[8] SAENZ J, AGELLI M. Investigating feature selection and random forests for inter-patient heartbeat classification[J]. Algorithms, 2020, 13(4): 75.
[9] GARCIA G, MOREIRA G, MENOTTI D, et al. Inter-patient ECG heartbeat classification with temporal VCG optimized by PSO[J]. Scientific Reports, 2017, 7(1): 10543.
[10] HASSAN A R, HAQUE M A. An expert system for automated identification of obstructive sleep apnea from single-lead ECG using random under sampling boosting[J]. Neurocomputing, 2017, 235: 122-130.
[11] KIRANYAZ S, INCE T, GABBOUJ M. Real-time patient-specific ECG classification by 1-D convolutional neural networks[J]. IEEE Transactions on Biomedical Engineering, 2015, 63(3): 664-675.
[12] RAJPURKAR P, HANNUN A Y, HAGHPANAHI M, et al. Cardiologist-level arrhythmia detection with convolutional neural networks[J]. arXiv:1707.01836, 2017.
[13] YILDIRIM ?. A novel wavelet sequences based on deep bidirectional LSTM network model for ECG signal classification[J]. Computers in Biology and Medicine, 2018, 96: 189-202.
[14] SAADATNEJAD S, OVEISI M, HASHEMI M. LSTM-based ECG classification for continuous monitoring on personal wearable devices[J]. IEEE Journal of Biomedical and Health Informatics, 2020, 24(2): 515-523.
[15] FENG K, PI X T, LIU H Y, et al. Myocardial infarction classification based on convolutional neural network and recurrent neural network[J]. Applied Sciences, 2019, 9(9): 1879.
[16] ZHAI X, TIN C. Automated ECG classification using dual heartbeat coupling based on convolutional neural network[J]. IEEE Access, 2018, 6: 27465-27472.
[17] JUN T J, NGUYEN H, KANG D, et al. ECG arrhythmia classification using a 2-D convolutional neural network[J]. arXiv:1804.06812, 2018.
[18] MURAT F, YILDIRIM ?, TALO M, et al. Application of deep learning techniques for heartbeats detection using ECG signals-analysis and review[J]. Computers in Biology and Medicine, 2020, 120: 103726.
[19] OH S L, NG E Y K, TAN R S, et al. Automated diagnosis of arrhythmia using combination of CNN and LSTM techniques with variable length heart beats[J]. Computers in Biology and Medicine, 2018, 96: 189-202.
[20] LI D, ZHANG J, ZHANG Q, et al. Classification of ECG signals based on 1D convolution neural network[C]//Proceedings of the 2017 IEEE 19th International Conference on e-Health Networking, Applications and Services (Healthcom), Dalian, Oct 12-15, 2017. Piscataway: IEEE, 2017: 1-6.
[21] JANGRA M, DHULL S K, SINGH K K. ECG arrhythmia classification using modified visual geometry group network (mVGGNet)[J]. Journal of Intelligent and Fuzzy Systems, 2020, 38(5): 1-15.
[22] MOUSAVI S, AFGHAH F. Inter- and intra- patient ECG heartbeat classification for arrhythmia detection: a sequence to sequence deep learning approach[C]//Proceedings of the 2019 IEEE International Conference on Acoustics, Speech and Signal Processing, Brighton, May 12-17, 2019. Piscataway: IEEE, 2019: 1308-1312.
[23] YILDIRIM ?, BALOGLU U B, TAN R S, et al. A new approach for arrhythmia classification using deep coded features and LSTM networks[J]. Computer Methods and Programs in Biomedicine, 2019, 176: 121-133.
[24] FERNANDEZ R, RENDEL A, RAMABHADRAN B, et al. Using deep bidirectional recurrent neural networks for prosodic-target prediction in a unit-selection text-to-speech system[C]//Proceedings of the 16th Annual Conference of the International Speech Communication Association, Dresden, Sep 6-10, 2015: 1606-1610.
[25] IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[J]. Computer Science, 2015, 37: 448-456.
[26] MISRA D. Mish: a self regularized non-monotonic activation function[J]. arXiv:1908.08681, 2019.
[27] ACHARYA U R, OH S L, HAGIWARA Y, et al. A deep convolutional neural network model to classify heartbeats[J]. Computers in Biology and Medicine, 2017, 89: 389-396.
[28] HUA X, HAN J, ZHAO C, et al. A novel method for ECG signal classification via one-dimensional convolutional neural network[J]. arXiv:2006.11655, 2020.
[29] BOUNY L, KHALIL M, ABDELLAH A. ECG heartbeat classification based on multi-scale wavelet convolutional neural networks[C]//Proceedings of the 2020 IEEE International Conference on Acoustics, Speech and Signal Processing, Barcelona, May 4-8, 2020. Piscataway: IEEE, 2020: 3212-3216.
[30] LI P, WANG Y, HE J, et al. High-performance personalized heartbeat classification model for long-term ECG signal[J]. IEEE Transactions on Biomedical Engineering, 2017, 64(1): 78-86.
[31] SINGH B N, TIWARI A K. Optimal selection of wavelet basis function applied to ECG signal denoising[J]. Digital Signal Processing, 2006, 16(3): 275-287.