Basis pursuit sparse decomposition using tunable-Q wavelet transform (BPSD-TQWT) for denoising of electrocardiograms

Area of Science:

• Biomedical Engineering
• Signal Processing

Background:

• Electrocardiogram (ECG) is crucial for diagnosing cardiac issues.
• Noise interference, specifically power line interference (PLI) and baseline drift (BD), degrades ECG signal quality.
• Effective noise reduction is essential for accurate cardiac abnormality identification.

Purpose of the Study:

• To propose a novel technique, basis pursuit sparse decomposition using tunable-Q wavelet transform (BPSD-TQWT), for removing PLI and BD from ECG recordings.
• To optimize BPSD-TQWT parameters (redundancy 'r' and decomposition levels 'J2') for effective noise suppression.
• To evaluate the performance of the proposed method against existing techniques.

Main Methods:

• Utilized tunable-Q wavelet transform (TQWT) for signal decomposition based on Quality factor (Q).
• Employed basis pursuit sparse decomposition (BPSD) to isolate and remove PLI and BD components.
• Optimized redundancy (r) and decomposition levels (J2) using signal-to-noise ratio (SNR) through trial-and-error.
• Validated the method on the MIT-BIH Arrhythmia database and real-time wearable ECG data.

Main Results:

• BPSD-TQWT significantly suppressed PLI with r=3, J2=10 and removed BD with r=4, J2=19.
• The method demonstrated improved performance metrics (SNR, MAX, NCC) compared to IIR filter, SWT, NLM, and LM methods.
• On the MIT-BIH database, SNR improved by 4.3 dB and 6.8 dB over IIR and SWT, respectively.
• On real-time data, SNR improved by 0.3 dB and 0.6 dB over IIR and SWT, respectively.

Conclusions:

• The proposed BPSD-TQWT method is highly effective for removing PLI and BD from ECG signals.
• BPSD-TQWT offers superior performance compared to conventional methods like IIR, SWT, NLM, and LM.
• This technique enhances ECG signal quality, leading to more reliable cardiac diagnostics.