Calculating heart rate from acceleration signals containing cardiac activity signals
A heart monitor includes a single chest accelerometer ( 210 ), an analog signal conditioning and sampling section ( 215 ) responsive to said accelerometer to produce a digital signal substantially representing acceleration, and a digital processor ( 220 ) operable to filter the acceleration signal into a signal affected by body motion and to cancel the body motion signal from the acceleration signal, thereby to produce an acceleration-based cardiac-related signal. Other processes and electronic systems are also disclosed.
1. A process of operating a data acquisition and signal processing unit comprising:
A. receiving acceleration signals that contain cardiac activity signals at an input of the unit;
B. digitally low pass filtering the acceleration signals with a filter cutoff of substantially a power line frequency to obtain filtered accelerometer signals;
C. decimating the filtered accelerometer signals by a factor of 10 to obtain decimated filtered accelerometer signals, x(i);
D. Savitzky-Golay polynomial smoothing filtering the decimated filtered accelerometer signals to obtain relatively slow-varying motion wander signals g(i);
E. subtracting the slow-varying motion wander signals g(i) from the decimated filtered accelerometer signals, x(i) to obtain heart sound signals S1 and S2, r(i);
F. generating envelope-processed residue signals R(i) from the heart sound signals S1 and S2, r(i);
G. performing a folded correlation process of the envelope-processed residue signals R(i) to obtain folded correlation signals fc(i);
H. locating peaks of the folded correlation signals fc(i) using an electronic amplitude-based peak picking process;
I. counting the peaks to calculate heart rate signals; and
J. providing calculated heart rate signals to an output of the unit.
2. The process of claim 1 in which digitally low pass filtering the acceleration signals includes filtering with a rolloff frequency less than 60 Hertz to attenuates 60 Hertz power line interference, and a rolloff frequency less than 50 Hertz to attenuate using 50 Hertz power line interference.
3. The process of claim 1 including processing the residue signals R(i) in overlapping frames indexed (i).
4. The process of claim 1 in which the Savitzky-Golay polynomial smoothing filtering uses 28th order and 401 point frame.
5. The process of claim 1 in which the folded correlation process includes folding heart monitoring residue samples R(i) from the later half of a frame around the center heart monitoring sample R(i) in the frame and multiplied by dot product with heart monitoring residue samples R(i) in the earlier half of the frame to obtain an instant (i) of an input residue signal stream R(i) in the center of the frame.
6. The process of claim 1 including performing at least some of the steps on a computer.