Self Tuned Adaptive Filter for Estimation and Removal of Power Line Interference from Electrocardiogram

Abstract

Electrocardiogram (ECG) becomes more vulnerable to external noises as compared to other biomedical signal and this is due to the non-invasive nature of ECG and the environment in which it is being recorded. Among other external noises, power line interference (PLI) is the most disquieting noise. In real world scenario, the problem becomes more complex due to drifts in fundamental frequency of PLI. Therefore, a high-resolution estimation of PLI frequency and its tracking is required for successful removal of PLI without affecting signal quality. Frequency estimation based on FFT, requires batch processing of data, which is not suitable for online application because of group delay. In this paper, a combination of State Space Recursive Least Square (SSRLS) adaptive filter and interpolation in time domain is employed for high resolution estimation and elimination of PLI frequency. Nevertheless, SSRLS is employed for adaptively tracking of sinusoidal component of PLI whose periodic time leads to frequency estimation in time domain; however, with low resolution depending upon sampling frequency of the signal. High resolution frequency estimation is achieved through time domain interpolation of peak segments of sinusoidal output of SSRLS with the advantage of reduced computational complexity. The performance of SSRLS is successively improved by tuning it with ongoing PLI frequency. The proposed self-tuned adaptive filter configuration is also capable of automatically tracking and removal the PLI irrespective of frequency of power system i.e, 50/60 Hz. The results show that proposed algorithm successfully removes the high-resolution frequency of PLI and tracks frequency drifts as well.