Motion-Artifact Reduction in Capacitive Heart-Rate Measurements by Adaptive Filtering

Electrocardiography (ECG) measurements are essential components in clinical diagnosis and monitoring. Conventional ECG measurements produce discomfort to the patients due to the use of gel and direct skin contact. Capacitive electrodes can measure ECG signals through an isolation layer; they are especially suitable for long-term ambulatory monitoring. However, capacitive ECG measurements are severely affected by motion artifacts (MAs) due to variable coupling distance. Adaptive filtering has been widely used for MA reduction in ECG measurements. Unfortunately, a reference signal recorded by additional sensors is required for the existing adaptive-filtering methods, limiting their applicability in ambulatory settings. In this paper on capacitive ECG recordings, a novel adaptive-filtering method is proposed for MA removal, where the reference signal is extracted from the power-line interference (PLI). PLI is particularly evident in a capacitive ECG due to unbalanced coupling capacitance. Along with MAs, electrode movement causes variations in the PLI amplitude. By demodulating the PLI, a reference signal reflecting variations in the coupling capacitance can be extracted for adaptive MA removal. The proposed method was evaluated both by simulations and real data and compared with the acceleration-based adaptive-filtering method. Comparable or higher ECG signal-to-noise ratio was achieved by the proposed method with a computational cost of 79 mu s/iteration, indicating effective MA removal. The proposed method may, therefore, lead to improved analysis of capacitive ECG signals in ambulatory settings.