Signal processing methods for non-contact cardiac detection using Doppler radar

Abstract

We investigate two signal processing methods, time and frequency-based approaches, for non-contact cardiac detection. A 5.8-GHz Doppler radar sensor was used for non-contact measurement of respiration and heart rate from stationary and semi-stationary subjects at a distance of 0.5 to 1.5 meters. We report on the accuracy of the heart rate measurements obtained using the two signal processing approaches, as compared to a reference heart rate obtained using a pulse oximeter. Simultaneous Photoplethysmograph (PPG) and non-contact sensor recordings were recorded over fifteen minute periods for five healthy subjects (4M/1F, ages 29.2±4.32 yrs) One algorithm is based on extraction of heart rate using spectral analysis and a second algorithm is based on automated detection of individual peaks associated with each cardiac cycle. Peaks were also extracted manually for comparison with the automated method. The peak-detection method was less accurate than the spectral method but suggests the possibility of acquiring beat by beat data. It was found that the signal SNR affects the ability to detect heart rates. An increase in SNR could lead to more accurate rates being obtained. The results described provide an estimate of the accuracy expected when measuring heart rate using a Doppler radar sensor. Non-contact measurement of heart rate will be useful in chronic disease monitoring for conditions such as heart failure and cardiovascular disease.