Pulse Wave Denoising Based on Improved Complementary Ensemble Empirical Mode Decomposition

Yong Chen; Zhimin Yao; Huanlin Liu; Junpeng Liao; Li Xu; Yanqing Feng

doi:10.3788/AOS231695

Acta Optica Sinica, Volume. 44, Issue 7, 0707001(2024)

Pulse Wave Denoising Based on Improved Complementary Ensemble Empirical Mode Decomposition

Yong Chen^1、*, Zhimin Yao¹, Huanlin Liu², Junpeng Liao¹, Li Xu¹, and Yanqing Feng¹

¹Key Laboratory of Industrial Internet of Things & Network Control, Ministry of Education, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

²School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(14)

Fig. 1. Pulse wave signal acquisition method

Download full size

Fig. 2. Simulated signal graph. (a) Signal with 5 dB Gaussian white noise; (b) signal with 10 dB Gaussian white noise; (c) signal with 15 dB Gaussian white noise; (d) signal with 20 dB Gaussian white noise

Download full size

Fig. 3. Decomposition diagram of 5 dB simulation signal

Download full size

Fig. 4. Mean mutual information of different white noise amplitudes for 5 dB simulation signal

Download full size

Fig. 5. Pulse wave denoising flow chart

Download full size

Fig. 6. FBG pulse wave detection device photo. (a) Wrist band; (b) pulse wave detection device

Download full size

Fig. 7. CSI algorithm for processing original pulse wave signal. (a) Original pulse wave signal; (b) processed pulse wave signal

Download full size

Fig. 8. Initial IMF component obtained by EEMD

Download full size

Fig. 9. Denoising effects of different algorithms. (a) CEEMD; (b) IIWT; (c) TIWT; (d) proposed algorithm

Download full size

Table 1. SNR results for different algorithms
View table
Table 1. SNR results for different algorithms
Original /dB CEEMD^［20］ IIWT^［11］ TIWT^［12］ Proposed
5 9.574 9.885 13.311 15.785
10 13.551 15.110 19.826 20.458
15 18.638 20.261 25.750 27.653
20 19.785 23.922 28.411 29.526
25 20.569 28.816 29.328 31.959

Table 2. RMSE results for different algorithms
View table
Table 2. RMSE results for different algorithms
Original /dB CEEMD^［20］ IIWT^［11］ TIWT^［12］ Proposed
5 2.913 2.719 1.833 1.251
10 2.045 1.491 0.865 0.793
15 1.523 0.823 0.437 0.305
20 0.978 0.541 0.322 0.296
25 0.746 0.308 0.291 0.215

Table 3. Basic physiological data of participants
View table
Table 3. Basic physiological data of participants
Characteristic Number or mean $\pm$ SD
Number n 5
Age /a 24.3 $\pm$ 2.1
Height /cm 173 $\pm$ 5.1
Weight /kg 72 $\pm$ 12.3
BMI /（kg/m²） 20.6 $\pm$ 4.2
Systolic/diastolic blood
pressure /mmHg
113.5/72.1 $\pm$ 17.2/5.8

Table 4. Mutual information mean values of IMF components corresponding to different white noise amplitudes
View table
Table 4. Mutual information mean values of IMF components corresponding to different white noise amplitudes
Amplitude of white noise Mean value of multual information
0.10 0.852
0.20 0.834
0.30 0.821
0.35 0.785
0.40 0.806

Table 5. Correlation coefficients between IMF components and original pulse wave signals
View table
Table 5. Correlation coefficients between IMF components and original pulse wave signals
IMF component Correlation coefficient IMF component Correlation coefficient
IMF1 0.2311 IMF7 0.0089
IMF2 0.8767 IMF8 0.0047
IMF3 0.8895 IMF9 0.0043
IMF4
IMF5
0.9234
0.0312
IMF10
IMF11
0.0035
0.0033
IMF6 0.0269 — —

Tools

Get Citation

Copy Citation Text

Yong Chen, Zhimin Yao, Huanlin Liu, Junpeng Liao, Li Xu, Yanqing Feng. Pulse Wave Denoising Based on Improved Complementary Ensemble Empirical Mode Decomposition[J]. Acta Optica Sinica, 2024, 44(7): 0707001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites