Hilbert Vibration Decomposition of Seismocardiogram for HR and HRV Estimation

Abstract

This paper presents a new time-varying decomposition method based on the Hilbert Vibration Decomposition (HVD) for estimating heart rate from a Seismocardiogram (SCG). The heart rate (HR) estimation method consists of signal decomposition using the HVD algorithm, heart rate envelope generation, and peak detection from the smooth envelope for beat-to-beat interval calculation. We derived the heart rate variability (HRV) metrics from the interbeat intervals. The method doesn’t require a reference ECG signal. The same signals are also subjected to Empirical Mode Decomposition (EMD) and Variational Mode Decomposition (VMD) methods. To compare these three decomposition methods, the CEBS database from Physionet Archive was used for testing and validation. The results show better accuracy in beat-to-beat interval estimation using the HVD method than others, and HRV metrics are accurately derived using our methodology. The performance results demonstrate that the standard ECG-derived heartbeats and HRV metrics and SCG-derived heartbeats and HRV metrics are comparable for healthy subjects.