Reduction of power line interference in electrocardiographic signals by dual Kalman filtering

Abstract

This paper presents a filter for reducing powerline interference in electrocardiographic signals (ECG), based on dual parameter and state estimation using with a Kalman filter. Two models were used to represent power-line interference and ECG signal. Both models were combined to simulate the ECG signal whose state was estimated for separating the ECG signal from the interference. The proposed algorithm was fine-tuned and compared using a set of tests relying on the QT arrhythmia database. Tuning tests were done for tracking clean ECG; these results were used for setting the algorithm’s parameters for later filtering tests. Exhaustive filtering tests were carried out on artificially corrupted database registers for given signal to noise ratios; performance curves were thus obtained, leading to comparing the proposed algorithm with other filtering methods. The proposed algorithm was compared to an recursive infinite impulse response filter (IIR) and a Kalman filter based on a simpler model. A filtering algorithm was thus obtained which is robust for changes in interference amplitude and keeps these properties for different types of ECG morphologies.