Improvement of power line model in ECG for interference reduction using EKF

Abstract

In this paper, we compare three different methods for suppressing sinusoidal perturbation, whose amplitude, frequency and phase are supposed to be unknown and may present changes along time, particularly, for removing power line interference from ECG signal records. The first method is a non-linear adaptive algorithm for extraction of non stationary sinusoids (NAENS), which is considered as a baseline algorithm of comparison. The second and third algorithms are based on Kalman filtering, using two different state space models of discrete-time sinusoidal oscillators. As a result, it was obtained that both Kalman filter algorithms have better performance than NAENS method, considering the presence of amplitude and frequency changes. © Springer-Verlag Berlin Heidelberg 2007.