The proposed bootstrap-based GWT method improves the accuracy of extracting heart and respiration rates from Doppler radar signals in noisy environments.
May enhance radar-based noncontact monitoring in lab settings; leaves open clinical validation before practice adoption.
In biomedical Doppler radar applications, the return signal is a nonlinear frequency-modulation (NLFM) random process whose phase conveys heart and respiration vital-sign information. These signatures modulate the phase of the signal as two oscillating components with frequencies less than a few hertz. Due to the nonstationary nature of these signals, their analysis by 1-D techniques, temporal and spectral, may not be very useful, and time-frequency techniques may be incapable of accurately extracting their instantaneous frequency (IF) trajectory. In this paper, we present a bootstrap-based generalized warblet transform (GWT) signal processing method. The presented signal processing tool is a parametric method that has a kernel with Fourier-series components. The coefficients of the kernel are estimated by an iteration procedure that converges to the IF of the radar signal. We show theoretically and experimentally that the bootstrap-based GWT can extract the amplitude and frequency of the two vital-sign components at a range of 3 m in the face of low signal-to-noise ratio and in the presence of phase noise and body motion artifacts, achieving an accuracy that is potentially better than conventional methods can provide.
No takes yet. Share an insight, caveat, or question.
Kazemi et al. (2015) studied this question.
Synapse has enriched 4 closely related papers on similar clinical questions. Consider them for comparative context: