Improved removal of electrocardiogram baseline wandering

Baseline wandering interference causes inaccurate detection of electrocardiogram (ECG) characteristics such as detection of the R point. This paper presents a novel phase space-based approach for baseline wandering removal of ECG signals. The baseline wander contaminated ECG signal is transformed to the phase space by using the signal derivative. Baseline wandering was eliminated by shifting loops of the ECG phase portrait on a predefined fixed position. This resulted in signal discontinuities in time domain which were solved using linear correction. The method was evaluated on synthetized ECG signals and on signals from the MIT-BIH database. Results show an improvement of signal-to-noise ratio after baseline wandering removal in phase space, comparable with related works, and efficient elimination of time-domain discontinuities in time domain.