Non-Contact Sensing of Seismocardiogram Signals Using Microwave Doppler Radar

Abstract

The seismocardiogram (SCG) signal is an effective representation of the heart contraction and blood ejection behavior for human subjects and exhibits in the vibrations of chest wall surface. The noncontact sensing of cardiac information using a microwave Doppler radar has the potential to monitor SCG without attaching sensors to the body. In this paper, the similarity of the radar acceleration waveforms (RAWs) to SCG is investigated in terms of the waveform morphology and fiducial points within a frequency band of 18–35 Hz, wherein the RAWs are less influenced by low frequency interference. A high morphological similarity is demonstrated by a high cross correlation coefficient between RAW and the dorso-ventral SCG (SCGz) waveform, including a coefficient over 0.9 for five of eight subjects and a minimum value of 0.72 for one subject experiencing interference. For the fiducial points, RAWs can correctly provide the locations of aortic valve opening (AO), an important fiducial point, with a root mean square deviation from the SCGz AO locations less than 2 ms for six of eight subjects. In addition, experiments are set up to evaluate the time shift of AO points before and after a 90-s exercise. The results show that RAWs can correctly determine the time shift directions of AO for all subjects, and estimate the decrease of pre-ejection period with an accuracy of less than 2 ms of that derived from SCG for six of eight subjects. These investigation results demonstrate the effectiveness of using the noncontact Doppler radar for SCG measurements.