Skin-Induced Motion Artifact Removal for Ambulatory Electrocardiography

Abstract

Skin-induced motion artifact (SIMA) is a troublesome problem for ambulatory electrocardiography (ECG) signal acquisition by e-skin dry sensor. Reference based algorithm is a promising way to tackle it. A highly correlated reference signal is the key for this solution. In this article, by using a new synthesized reference signal for LMS algorithm, which combines both skin induced impedance behavior and skin “slope” and recovery “time constant”, the SIMA related noise is greatly reduced. Firstly, we show the phenomenon that the SIMA “slope” and “recovery time constant” have some relationships with skin induced impedance, its derivatives and specific tissue characteristics. Secondly, We do the offline “slope” and “time constant” calibration work first, then measure the real time skin induced impedance, and then combine the above information together, to form a virtual reference signal, then we show the new reference has high correlation with actual SIMA noise. Finally, considering that SIMA potential consists of fast changing part and slow changing part, we use an improved LMS algorithm to handle the contradiction between speed and accuracy, which includes fast tracking part and low convergence error part. We set up the SIMA removal system in Matlab, and use actual motion artifact contaminated ECG signal as the input, and then do offline signal processing. Results show that SIMA is improved around <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$7\sim 15$ </tex-math></inline-formula> dB typically on average. Integrating skin-induced impedance with human body related “slope” and “time constant” as a reference signal seems a feasible method to remove skin-induced motion artifact.