Utilizing SimEMG to Evaluate the Effectiveness of Adaptive Filters in Removing EMG Contamination from ECG Recordings

The noise in electrocardiographic (ECG) recordings can significantly affect their medical interpretation, especially when the electromyographic (EMG) noise overlaps with the QRS complex, making it difficult to remove. Typically, noise-removal methods are evaluated using algorithms tested on artificially contaminated signals created by adding noise to noise-free ECG recordings. In this context, the SimEMG database containing EMG-noise-free and EMG-contaminated ECG signals is used. The paper uses SimEMG database to evaluate the effectiveness of adaptive filters based on an adaptive noise cancellation in removing real EMG contaminants, as well as other noise artifacts, from the ECG recorded signals. The evaluation is made by computing the Signal to Noise Ratio (SNR) of the ECG signal compared to the contaminant EMG and other noise before and after adaptive filtering. Different filtering algorithms, step sizes, and filter lengths are investigated. The obtained results indicate the effectiveness of the used method in removing the EMG artifacts and enhancing the ECG signal, achieving an improvement of above 3 dB at a certain value of the filter length and step size.