Respiratory rate estimation from multi-lead ECGs using an adaptive frequency tracking algorithm

Estimating the respiratory rate (RR) from the electrocardiogram (ECG) is of interest as the direct measurement of the respiration in clinical situations is often cumbersome. In this study, the RR was estimated from the multi-lead ECG R-peak amplitude (RPA) waveforms, which contain the modulation of the cardiac activity by the respiration. An adaptive oscillator-based frequency tracking algorithm was used to estimate the RR from the RPAs of two or three ECG leads. This automatic and instantaneous method tracks the common respiratory frequency which is present in its inputs as the RR estimate. On a subset of the Physionet MFH/MF dataset, it was shown that combining information from three leads yielded more accurate RR estimates than using two leads or each lead alone. It was also shown that the frequency tracking algorithm outperformed Fourier-based frequency estimation.