A Case Study for Removing Cardiac Artifacts from Non-Invasive Electrospinography Signals

The recording of spinal cord neuronal activity through high-density surface electromyography is susceptible to electrocardiographic artifacts, needed to be removed before further analysis. In this study, seven algorithms were compared: Template Subtraction, Adaptive Template Subtraction, High-pass filter with 200 Hz cutoff frequency, Adaptive Template Sub-traction and subsequent high-pass filtering, Stationary Wavelet Transform, Second order Extended Kalman Smoother, and Empirical Mode Decomposition. These algorithms were assessed using five performance indicators: Relative Error, Signal-to-Noise Ratio, Cross-Correlation, Spectral Distortion, and Kurtosis Ratio. Among the algorithms, Adaptive Template Subtraction demon-strated superior efficacy in denoising cardiac waves, regardless of the electrocardiographic artifact morphology. Despite potential limitations in fully preserving low-frequency components, its prowess in signal preservation renders it a valuable asset for future research endeavors spanning a wide frequency spectrum.