Accelerometer based motion noise analysis of ECG signal

Abstract

Continuous monitoring of an ECG signal is an important method to evaluate the subject's health state. However, there are many distortion factors in ECG signals, such as motion, respiration, and 60 Hz power noise, and EMI noise. Above all, motion is the most important factor in ECG signal distortion of moving subject and the baseline drift not only causes the estimation error of R-R interval, but changes ST segment level and results in diagnostic errors. The baseline drift removal algorithms without original ECG signal degradation are difficult to develop, because they affect the low frequency components in general. The accurate evaluation of ECG baseline drift must precede its removal. The main cause of baseline drift is the subject's motion, which can be evaluated by using small 3-axial accelerometer with accuracy and convenience. In this paper, baseline drift in ECG signal caused by motion artifact was quantitatively analyzed and evaluated by using 3-axial accelerometer and classifying gesture, posture, and dynamic motion. The baseline drift was influenced by respiration in static state and motion in dynamic state and the correlation between ECG and acceleration signal can be identified. The low frequency component of ECG baseline drift and high frequency component of acceleration signal were increased by motion and the facts were verified by correlation analysis. These results can be used to minimize the analysis error caused by motion. In addition, to apply baseline drift removal algorithm or not can be decided by the motion detection and the diagnostic error caused by the loss of low frequency component can be minimized. When this result is applied to the portable ECG device with a built-in accelerometer, subject's motion context information can be known and used to manage emergent situation.