Denoising Non-Invasive Electroespinography Signals by Different Cardiac Artifact Removal Algorithms

Area of Science:

• Biomedical Engineering
• Neuroscience
• Signal Processing

Background:

• Non-invasive spinal cord neuronal activity recording via electrospinography (ESG) using high-density surface electromyography (HD-sEMG) shows promise.
• Electrocardiographic (ECG) artifacts frequently contaminate ESG signals, necessitating effective denoising techniques.

Purpose of the Study:

• To evaluate seven established electromyography (EMG) denoising algorithms for their efficacy in removing ECG artifacts from ESG data.
• To assess the algorithms' ability to preserve the broad spectral bandwidth crucial for ESG analysis.

Main Methods:

• Seven denoising algorithms were tested: Template Subtraction (TS), Adaptive Template Subtraction (ATS), High-Pass Filtering at 200 Hz (HP200), ATS+HP200, Second-Order Extended Kalman Smoother (EKS2), Stationary Wavelet Transform (SWT), and Empirical Mode Decomposition (EMD).
• Performance was quantified using Relative Error (RE), Signal-to-Noise Ratio (SNR), Cross-Correlation (CC), Spectral Distortion (SD), Kurtosis Ratio (KR2), and ΔKR2.
• ESG data were collected from nine participants at brachial and lumbar plexus sites with varying electrode configurations.

Main Results:

• Adaptive Template Subtraction (ATS) demonstrated superior performance in suppressing diverse ECG artifact shapes compared to all other methods.
• ATS achieved the optimal balance between artifact removal and signal integrity, despite not fully preserving low-frequency components.
• Algorithm performance was enhanced in recordings with lower ECG contamination, particularly brachial plexus recordings with closely spaced reference electrodes.

Conclusions:

• ATS is the most effective algorithm for removing ECG artifacts in ESG recordings, providing a good balance between artifact suppression and signal preservation.
• The effectiveness of denoising algorithms is influenced by the level of ECG contamination and electrode configuration.
• Further research into ESG signal processing is warranted, with ATS serving as a strong baseline method.