Session P77.2

Improvement of an Extended Kalman Filter Power Line Interference Suppressor for ECG Signals

LD Avendaño Valencia*, LE Avendaño, CG Castellanos, JM Ferrero

Universidad Nacional de Colombia
Manizales, Columbia

A methodology to reduce power line interference from electrocardiographic (ECG) records is proposed, in order to track its amplitude, frequency and phase variations, without generation of distortions on the ECG signal. This methodology considers the non-stationary characteristic of the ECG and the power line interference. The methodology consists on three stages. First, the spectral components where the interference lies are separated from the ECG components by means of a second order IIR filter. Then, the specific 60Hz component is estimated using the extended Kalman filter, fed with a discrete sinusoidal oscillator. This estimator can track the variability of amplitude, phase and frequency of the power line interference. Once the interference is estimated, it is removed from the ECG signal, thus eliminating the power line interference. The Kalman filter parameters are tuned using a genetic algorithm with exponential ranking fitness scaling, which uses as evaluation function the correlation index mean for the QT database. For each point on the parameter space, filtering of artificially contaminated ECG registers is done. The filter’s evaluation surface reflects optimal performance on the parameter space for tested signal to noise ratios. As result, the filter’s performance reaches mean correlation index over 0.99 with variance lower than 1e-10, showing that the proposed methodology effectively reduces the power line interference, but preserving the ECG characteristics, while tracks the variability of amplitude, frequency and phase of the disturbance.

(Abstract Control Number: 215)