A two-step pre-processing tool to remove Gaussian and ectopic noise for heart rate variability analysis

Figures

• 

  Figure 1. Time-frequecy domain for CEEMDAN-WD denoising (a) Waveform of original artificial ECG signal, (b) Spectrogram of original artificial ECG, (c) Waveform of artificial ECG signal with 2 dB of white Gaussian noise, (d) Spectrogram of artificial ECG signal with 2 dB of white Gaussian noise, (e) Waveform of partially reconstructed denoised ECG signal by CEEMDAN-WD proposed method, (f) Waveform of partially reconstructed denoised ECG signal by CEEMDAN-WD proposed method.

• 

  Figure 2. Performance indicators for CEEMDAN-WD (a), Mean correlation coefficient for CEEMDAN-WD denoising algorithm at different levels of Gaussian noise, (b), Mean RMSE for CEEMDAN-WD denoising algorithm at different levels of Gaussian noise.

• 

  Figure 3. Comparison of different ectopic filters, (a) Artificial RR interval with 2% of ectopic beats, (b) Denoised RR interval by SDROM, (c) Denoised RR interval by ADF, (d) Denoised RR interval by combination filter SDROM-ADF, (e) Denoised RR interval by combination filter ADF-SDROM.

• 

  Figure 4. Comparison of performance indicators for ectopic filters (a), Mean correlation coefficient for SDROM, ADF, SDROM-ADF, and ADF-SDROM denoising algorithm at different levels of ectopic noise, (b), Mean RMSE for SDROM, ADF, SDROM-ADF, and ADF-SDROM denoising algorithm at different levels of ectopic noise.

• 

  Figure 5. Proposed method applied to real ECG signals, (a) Denoising of the ‘16,539’ ECG recording from the MIT-BIH normal sinus rhythm database, CEEMDAN-WD filtering Gaussian noise (left), and SDROM-ADF filtering ectopic beats (right), (b) Denoising of the ‘119’ ECG recording from the MIT-BIH Arrhythmia database, CEEMDAN-WD filtering Gaussian noise (left) and SDROM-ADF filtering ectopic beats (right), (c) Denoising of the ‘52’ ECG recording from the sudden cardiac death database, CEEMDAN-WD filtering Gaussian noise (left) and SDROM-ADF filtering ectopic beats (right).

• 

  Figure 6. Heat maps for −log10 transformed p-values for twenty-four of the most common HRV measures for real and artificial signals, (a) Heat map of −log10(p-value) of original clean and noisy artificial HRV measures for different levels of ectopic and Gaussian noise. The last column shows the artificial HRV measures for both types of noise combined at 6 dB of Gaussian and 6% of ectopic noise, (b) Heat map of −log10(p-value) of noisy and denoised (by the proposed algorithm) real HRV measures obtained from three Physionet databases: MIT-BIH Arrhythmia, MIT-BIH Normal Sinus Rhythm, and Sudden Cardiac Death database.

• 

  Figure 7. Examples of HRV measures that did not show a linear relationship in the relative change of HRV measure (y axis) with an increasing percentage of ectopic noise (x axis), (a) HF-Norm, (b) alpha 1, (c) Sample entropy.

• 

  Figure 8. Examples of HRV measures that did not show a linear relationship in the relative change of HRV measure (y axis) with incrementing Gaussian noise (x axis), (a) HF-Norm, (b) alpha 1, (c) Sample entropy.