Journal of Jilin University(Engineering and Technology Edition) ›› 2020, Vol. 50 ›› Issue (5): 1818-1825.doi: 10.13229/j.cnki.jdxbgxb20190504

Previous Articles    

Pulse wave signal classification algorithm based on time⁃frequency domain feature aliasing using convolutional neural network

Guo-hua LIU1,2(),Wen-bin ZHOU1,2   

  1. 1.School of Electronic Information and Optical Engineering, Nankai University, Tianjin 300350, China
    2.Key Laboratory of Photoelectric Sensors and Sensor Network Technology, Nankai University, Tianjin 300350, China
  • Received:2019-05-22 Online:2020-09-01 Published:2020-09-16

RICH HTML

 13   

PDF (PC)

956

Abstract:

A pulse wave classification algorithm of low complexity based on aliasing of time-frequency domain feature was proposed to solve the problems of low recognition rate and complex implementation in current research of pulse wave signal recognition. First, the time domain features of the pulse wave signal are extracted based on the Convolutional Neural Network(CNN), including the single-period feature characterizing the features of the signal segments in the period and the multi-period features characterizing the relationship between the cycles. Then the features in the frequency domain are expressed by the Mel cepstrum coefficients based on the wavelet transform. Finally, based on the aliasing and redundancy elimination of the time-frequency domain features using the neural network full connection layer, the pulse wave signal classification is realized by the Softmax classifier. The method can achieve feature extraction through low computational cost, due to the weight sharing and dimensionality reduction of CNN. In the simulation experiment based on Python platform, the recognition accuracy of the proposed method can reach 93%, which is much higher than the accuracy of traditional recognition algorithms based on time domain features or frequency domain features.

Key words: information processing technology, classification of pulse wave signal, convolutional neural network, aliasing of time-frequency domain features, wavelet transform

CLC Number: 

  • TN911.7

Fig.1

Extraction module of single-cycle feature"

Table 1

Parameter configuration of pulse wave signal single-cycle feature extraction module"

卷积模块1参数卷积模块2参数
卷积层卷积核维度7*1*1卷积层卷积核维度5*1*32
卷积核步长3*1*1卷积核步长2*1*1
卷积核个数32卷积核个数64
激活函数ReLU激活函数ReLU
池化层池化方式最大池化池化层池化方式最大池化
卷积核维度3*1*1卷积核维度2*1*1
卷积核步长2*1*1卷积核步长1*1*1

Fig.2

Extraction module of multi-cycle feature"

Table 2

Parameter configuration of pulse wave signal multi-cycle feature extraction module"

卷积模块3参数卷积模块4参数
卷积层卷积核维度7*3*1卷积层卷积核维度5*3*32
卷积核步长3*1*1卷积核步长2*1*1
卷积核个数32卷积核个数64
激活函数ReLU激活函数ReLU
池化层池化方式最大池化池化层池化方式最大池化
卷积核维度3*1*1卷积核维度2*1*1
卷积核步长2*1*1卷积核步长1*1*1

Fig.3

Frequency domain feature extraction module"

Fig.4

Algorithm structure of pulse wave classification"

Table 3

Parameter configuration of classification network"

网络结构参 数
神经元个数(全连接层1)128
神经元个数(全连接层2)64
激活函数ReLU函数
神经元个数(输出层)K
激活函数(输出层)Softmax函数
交叉验证算法留出法
代价函数交叉熵
网络更新算法动量梯度下降

Fig.5

Waveform comparison before and after pulse wave noise reduction"

Table 4

Test results of training set, test set, and verification set for classification network"

项目识别正确数识别错误数识别准确率
训练集407250.9421
验证集134100.9306
测试集13590.9375

Table 5

effect of different characteristics on classification accuracy of pulse wave signals"

项目识别正确数识别错误数识别准确率
单周期特征121230.8403
多周期特征125190.8681
单周期与多周期混合特征130140.9028
基于小波分析的MFCC75690.5208
时频域混叠特征13590.9375

Table 6

Performance comparison of this method with other classification methods"

方法识别准确率输入维度
本文0.937 53 042
文献[8]0.838 117 000

Fig.6

Influence of number of data sets in training set on recognition accuracy"

Table 7

Effect of different activation functions on performance of classification networks"

项目ReLUSigmoidTanh
训练集识别率0.94210.87960.8750
测试集识别率0.93750.87500.8681
训练时间(100次)/s27.430.729.12
迭代次数8431875996
1 郭维, 刘光达, 张晓枫, 等. 基于多特征参数综合分析脉搏波信号失真度算法的实现[J]. 吉林大学学报: 工学版, 2012, 42(4): 1044-1048.
Guo Wei, Li Guang-da, Zhang Xiao-feng, et al. Realization of algorithm of abnormal pulse wave signal detection based on associate factor of pulse features[J]. Journal of Jilin University(Engineering and Technology Edition), 2012, 42(4): 1044-1048.
2 张爱华, 王平, 丑永新. 基于动态差分阈值的脉搏信号峰值检测算法[J]. 吉林大学学报: 工学版, 2014, 44(3): 847-853.
Zhang Ai-hua, Wang Ping, Yong-xin Chou. Peak detection of pulse signal based on dynamic difference threshold[J]. Journal of Jilin University(Engineering and Technology Edition), 2014, 44(3): 847-853.
3 马佳辉. 基于脉搏信号的人体脉搏特征参数的研究[D].北京： 北京邮电大学电子工程学院, 2018.
Ma Jia-hui. Research of pulse characteristic parameters of human body based on pulse signal[D]. Beijing: School of Electronic Engineering, Beijing University of Posts and Telecommunications, 2018.
4 杜昭慧, 司玉娟. 基于层次支持向量机的脉搏信号情感识别[J]. 吉林大学学报: 信息科学版, 2017, 35(1): 37-42.
Du Zhao-hui, Si Yu-juan. Emotion recognition of pulse signal based on hierarchical support vector machine[J]. Journal of Jilin University(Information Science Edition), 2017, 35(1): 37-42.
5 柳兆荣, 李惜惜. 关于脉图的分析[J]. 中国科学(B辑), 1983, 13(2): 131-138.
Liu Zhao-rong, Li Xi-xi. Analysis of the pulse map[J]. Chinese Science(Series B), 1983, 13(2): 131-138.
6 唐铭一, 李凯, 马小铁. 脉搏波信号时域特征提取与算法的研究[J]. 计算机与现代化, 2010(4): 15-17, 22.
Tang Ming-yi, Li Kai, Ma Xiao-tie. Research on pulse wave signal and time-domain feature extraction algorithm[J]. Computer and Modernization, 2010(4): 15-17, 22.
7 徐可欣, 王继寸, 余辉, 等. 脉搏波时域特征与血压相关性的研究[J]. 中国医疗设备, 2009, 24(8): 42-45.
Xu Ke-xin, Wang Ji-cun, Yu Hui, et al. Research of correlation between time-domain characteristics of the pulse wave and blood pressure[J]. China Medical Equipment, 2009, 24(8): 42-45.
8 胡阳生. 基于卷积神经网络的脉搏分析方法研究[D]. 北京：北京邮电大学电子工程学院, 2018.
Hu Yang-sheng. Study on pulse analysis method based on convolution neural network[D]. Beijing: School of Electronic Engineering, Beijing University of Posts and Telecommunications, 2018.
9 陈星池, 黄淑春, 赵海, 等. 基于极限学习机的中医脉象识别方法[J]. 东北大学学报: 自然科学版, 2017, 38(9): 1226-1229.
Chen Xing-chi, Huang Shu-chun, Zhao Hai, et al. Recognition method of traditional Chinese medicine pulse conditions based on extreme learning machine[J]. Journal of Northeastern University(Natural Science), 2017, 38(9): 1226-1229.
10 董丽娜, 何怡, 叶卫平. 基于小波分析的梅尔频率倒谱参数[J]. 北京师范大学学报: 自然科学版, 2015, 51(5): 469 -474.
Dong Li-na, He Yi, Ye Wei-ping. Wavelet analysis based Mel frequency cepstrum parameters[J]. Journal of Beijing Normal University(Natural Science), 2015, 51(5): 469-474.
11 Nogueira D M, Ferreira C A, Jorge A M. Classifying heart sounds using images of MFCC and temporal features[C]∥Portuguese Conference on Artificial Intelligence, Springer, Cham, 2017: 186-203.
12 高印寒, 谢军, 梁杰, 等. 基于小波分析的听觉滤波器组模型[J]. 吉林大学学报: 工学版, 2008， 38(): 177-181.
Gao Yin-han, Xie Jun, Liang Jie, et al. Auditory filter bank model based on wavelet transform[J]. Journal of Jilin University(Engineering and Technology Edition), 2008, 38(Sup.1): 177-181.
13 Goldberger A L, Amaral L A N, Glass L, et al. PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals[J]. Circulation, 2000, 101(23): 215-220.
[1] Zai-feng SHI,Jin-zhuo LI,Qing-jie CAO,Hui-long LI,Qi-xing HU. Low⁃dose spectral computer⁃tomography imaging denoising method via a generative adversarial network [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1755-1764.
[2] Ke-yan WANG,Di WANG,Xi ZHAO,Jing-yi CHEN,Yun-song LI. Image dehazing based on joint estimation via convolutional neural network [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1771-1777.
[3] Xiang-jiu CHE,You-zheng DONG. Improved image recognition algorithm based on multi⁃scale information fusion [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1747-1754.
[4] Hua CHEN,Wei GUO,Jing-wen YAN,Wen-hao ZHUO,Liang-bin WU. A new deep learning method for roads recognition from SAR images [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(5): 1778-1787.
[5] Wei ZHANG,Yong HAN,Ming JIN,Xiao-lin QIAO. Toeplitz matrices reconstruction based DOA estimation for coherent signals [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(2): 703-710.
[6] Yan-fen CHENG,Li-juan YAO,Qiao YUAN,Xian-qiao CHEN. Clearing strategy of underwater video image [J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(2): 668-677.
[7] Xiao-hui YU,Zhi-cheng ZHANG,Xin-bo LI,Xiao-dong SUN. Parameter estimation of exponentially damped sinusoidsbased on state⁃space model [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 2083-2088.
[8] Xiang-jiu CHE,Hua-luo LIU,Qing-bin SHAO. Fabric defect recognition algorithm based onimproved Fast RCNN [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 2038-2044.
[9] Fu LIU, Mei-jing QUAN, Ke WANG, Yun LIU, Bing KANG, Zhi-wu HAN, Tao HOU. Indoor positioning method based on location fingerprinting of imitating mechanism of scorpion vibration source [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 2076-2082.
[10] Hong-wei ZHAO,Peng WANG,Li-li FAN,Huang-shui HU,Ping-ping LIU. Similarity retention instance retrieval method [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 2045-2050.
[11] Zi-ji MA,Hao LU,Yan-ru DONG. Dual path convolutional neural network forsingle image super⁃resolution [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(6): 2089-2097.
[12] Yang LU,Shi-gang WANG,Wen-ting ZHAO,Yan ZHAO. Facial expression recognition based on separability assessment of discrete Shearlet transform [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(5): 1715-1725.
[13] Chao DONG,Jing⁃hong LIU,Fang XU,Ren⁃hao WANG. Fast ship detection in optical remote sensing images [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(4): 1369-1376.
[14] Ke⁃yan WANG,Yan HU,Huai WANG,Yun⁃song LI. Image dehazing algorithm by sky segmentation and superpixel⁃level dark channel [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(4): 1377-1384.
[15] Xiong-fei LI,Lu SONG,Xiao-li ZHANG. Remote sensing image fusion based on cooperative empirical wavelet transform [J]. Journal of Jilin University(Engineering and Technology Edition), 2019, 49(4): 1307-1319.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
Copyright © Journal of Jilin University(Engineering and Technology Edition), All Rights Reserved.
Tel: +86-431-85095297　Fax: +86-431-85094074　E-mail: xbgxb@jlu.edu.cn
Powered by Beijing Magtech Co. Ltd