Electrocardiogram baseline wander removal based on empirical mode decomposition

In electrocardiogram (ECG) signals, low frequency artifacts caused by respiration and movements of patient and recording hardware, create challenging problems for signal processing algorithms. Due to nonlinear origins of these undesirable artifacts, the use of nonlinear signal processing methods gives assistance to achieve more reliable outcomes. This paper introduces a baseline drift cancellation algorithm using polynomial fitting based on empirical mode decomposition (EMD), then evaluates it comparing two conventional methods, EMD-based global slope minimization and moving average filtering. The methods were applied over 26 generated ECG baselines where the polynomial fitting showed more than 81% correlation with the generated baselines. The comparative results proved a remarkable robustness of the proposed algorithm against the baseline drifts.