[1]逯鹏,王汉章,毛晓波,等.基于卷积自编码器网络的脉搏波分类模型[J].郑州大学学报(工学版),2021,42(5):56-61.[doi:10.13705/j.issn.1671-6833.2021.05.004]
　LU Peng,WANG Hanzhang,MAO Xiaobo,et al.Pulse Wave Classification Model Based on Convolutional Autoencoder[J].Journal of Zhengzhou University (Engineering Science),2021,42(5):56-61.[doi:10.13705/j.issn.1671-6833.2021.05.004]
点击复制

基于卷积自编码器网络的脉搏波分类模型()
分享到：

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

卷:
42
期数:
2021年5期
页码:
56-61
栏目:
出版日期:
2021-09-10

文章信息/Info

Title:
Pulse Wave Classification Model Based on Convolutional Autoencoder
作者:
逯鹏1,2,王汉章1,2,毛晓波1,2,赵宇平2,3,刘超1,2,尚莉伽4,孙智霞5
1.郑州大学 电气工程学院，河南 郑 州 450001；2.中 医 药 智 能 科 学 与 工 程 技 术 研 究 中 心，河 南 郑 州 450001；3.中国中医科学院，北京 100020；4.北京市东城区中小学卫生保健所，北京 100007；5.郑州大学第五附属医院，河南 郑州 450052
Author(s):
LU Peng1,2, WANG Hanzhang1,2, MAO Xiaobo1,2, ZHAO Yuping2,3, LIU Chao1,2, SHANG Lijia4, SUN Zhixia5
1.Institute of Electric al Engineering， Zhengzhou University， Zhengzhou 450001， China； 2.Research Center for Intelligent Science and Engineering Technology of TCM, Zhengzhou 450001， China； 3.China Ac ademy of Chinese Medic al Sciences， Beijng 100020， China； 4.Primary and Secondary School Health Care in Beijing Dongcheng District， Beijing 100007， China； 5.The Fifth Affiliated Hospital of Zhengzhou University， Zhengzhou 450052， China
关键词:
Keywords:
cardiovascular disease;  pulse wave;  convolutional neural network;  autoencoder
DOI:
10.13705/j.issn.1671-6833.2021.05.004
文献标志码:
A
摘要:
针对基于深度学习的脉搏波分类依赖大量标记数据,临床数据有限、标注成本高影响了脉搏波分类识别效果,设计一种基于卷积自编码器网络( CAE-Net) 的脉搏波分类模型。 首先,利用卷积神经网络( CNN)的局部特征提取能力和自编码器(AE )的压缩重构及降维特性构建卷积自编码器( CAE ) ,结合脉搏波波形特点,在平均绝对误差损失函数中引入脉搏波时域特征约束,提升 CAE 对脉搏波低维特征自学习能力；其次,重用预训练 CAE 编码层网络和权重,构建 CAE-Net,并利用有标记脉搏波数据进行网络微 调。 在 心 血 管 疾 病 数 据 集 上 的 实 验 结 果 表 明, CAE-Net 的 分 类 准 确 率 为 98.00%, F1 值 达 到94.40%,相较于其他脉搏波分类模型,设计网络所提取特征的区分度较高,同时减弱了对标注脉搏波数据的依赖,在小样本脉搏波数据上具有较好分类效果。
Abstract:
Classification of pulse wave based on deep learning relies on a large number of labeled data, however, limited clinical data and expensive labeling costs hinder the pulse wave classification and recognition. A pulse wave classification model based on convolutional autoencoder networks (CAE-Net) is designed in this paper. Firstly, the convolutional autoencoder (CAE) is constructed, which combines the local feature extraction ability of convolutional neural network (CNN) and the compression reconstruction and dimension reduction characteristics of autoencoder (AE). And considering the characteristics of pulse wave, the time domain feature constraint of pulse wave is introduced into the mean absolute error loss function to improve the self-learning ability of CAE for low dimensional features. Secondly, the CAE-Net is constructed by reusing the coding layer network and weights of the pre-training CAE, then the network is fine tuned by using labeled pulse waves. Experiments on cardiovascular disease dataset show that the classification accuracy of CAE-Net is 98.00%, and the F1 score is 94.40%. Compared with other classification models, the designed network can extract features with high discrimination, reduce the dependence on the labeled pulse waves, and perform well in the classification of small sample pulse wave data.

参考文献/References:

[1] LIU W C,FANG X,CHEN Q Q,et al.Reliability analysis of an integrated device of ECG, PPG and pressure pulse wave for cardiovascular disease[J].Microelectronics reliability,2018,87:183-187.

[2] IQBAL U,YING WAH T,HABIB UR REHMAN M,et al.Usage of model driven environment for the classification of ECG features:a systematic review[J].IEEE access,2018,6:23120-23136.
[3] KOREN A，JUR CEVI CM，PRASAD R.Comparison of data-driven models for cleaning eHealth sensor data：use c ase on ECG signal[J].Wireless personal communications，2020，114（2） ：1501-1517.
[4] SHARMA H.Heart rate extraction from PPG signals using variational mode decomposition [J]. Biocybernetics and biomedic al engineering， 2019， 39（1） ：75-86.
[5] 刘国华，周文斌.基于卷积神经网络的脉搏波时频域特征混叠分类[J].吉林大学学报 （工学版），2020，50（5）：1818-1825.
[6] ISLAM M T，ZABIR I，AHAMED S T，et al.A time-frequency domain approach of heart rate estimation from photoplethysmographic（PPG）signal [J].Biomedical signal processing and control，2017，36：146-154.
[7] HASSANI A， FORUZAN A H.Improved PPG-based estimation of the blood pressure using latent space features[J].Signal，image and video processing，2019，13（6） ：1141-1147.
[8] LI G Y，WATANABE K，ANZAI H，et al.Pulse-wave pattern classific ation with a convolutional neural network[ J].Scientific reports，2019，9：14930.
[9] KWON S， HONG J， CHOI E K， et al.Deep learning approaches to detect atrial fibrillation using photoplethysmographic signals： algorithms development study[J].JMIR mHealth and uHealth，2019，7（6） ：e12770.
[10] HUANG Z W， XUE W T， MAO Q R， et al. Unsupervised domain adaptation for speech emotion recognition using PCANet [J].Multimedia tools and applic ations，2017，76（5） ：6785-6799.
[11] CHEN M， SHI X B， ZHANG Y， et al.Deep features learning for medic al image analysis with convolutional autoencoder neural network [ J ].IEEE transactions on big data， 2017:1-1.
[12] YANG J L， BAI Y， LIN F， et al.A novel electroc ardiogram arrhythmia classific ation method based on stacked sparse auto-encoders and softmax regression[J].International journal of machine learning and cybernetics，2018，9（10） ： 1733-1740.
[13] 逯鹏，李奇航，尚莉伽，等.基于优化极限学习机的CVD 预测模型研究 [J].郑州大学学报（工学版），2019，40（2） ：1-6.
[14] 季忠，刘旭.基于波形特征和小波的脉搏波特征点识别研究[J].仪器仪表学报，2016，37（2） ：379-386.
[15] ZHANG Z C， ZHANG Y， YAO L N， et al.A sensor based wrist pulse signal processing and lung cancer recognition [ J ].Journal of biomedic al informatics，2018，79：107-116.
[16] 张选，胡晓娟.基于GoogLeNet和ResNet的深度融合神经网络在脉搏波识别中的应用[J].计算机系统应用，2019，28（10） ：15-26.

更新日期/Last Update: 2021-10-11