Dynamic Segmented Beat Modulation Method for Denoising ECG Data from Wearable Sensors

Wearable electrocardiogram (ECG) sensing devices are defining the future of continuous, long-term and automatic cardiac health monitoring. The target problem for ECG acquisition and processing through patch-based wearable sensors is the noise induced by motion artifacts during daily life activities. In this work we propose an improvement of segmented beat modulation method (SBMM) with dynamic template to denoise ECG record acquired from a patch-based ECG sensor tested on healthy subjects while performing different activities: drinking coffee, typing keyboard, pressing and releasing signal electrode, walking at 1 and 3 miles per hour and running at 5 and 7 miles per hour inducing diverse motion artifacts. The robustness of the proposed algorithm is reported in terms of improvement in signal-to-noise ratio (SNR) corresponding to high, medium and low levels of input signal noise. Results stratified by daily activities indicate that dynamic-template SBMM filtering yields an overall improvement in SNR for ECG signals corrupted by motion artifacts supporting the hypothesis that the dynamic-template SBMM is an efficient denoising algorithm for ECG signal processing acquired through wearable sensors.