Bed-Embedded Heart and Respiration Rates Detection by Longitudinal Ballistocardiography and Pattern Recognition

Figure 1. A Diagram of the commonest form of the normal ballistocardiogram with the letters denoting its various waves (tips), redrawn after Starr’s publication [6].

Figure 2. The Load cell setup to measure the body forces in the transverse/dorsoventral axis for comparison with the signal in the longitudinal axis: (a) The load cell structure diagram; (b) the actual sensor placement relative to the bed; the mattress is lifted temporarily for the sake of the picture.

Figure 3. The load cell setup for the longitudinal BCG: (a) The load cell structure diagram; (b) the complete mechanical structure; (c) the structure fitted at the top side of the bed base. The clamps fit on the bed base from underneath.

Figure 4. The complete measurement system.

Figure 5. Frequency spectrum diagrams of raw BCG signals taken from two volunteers, both lying in the supine position. The major signal components are annotated. The upper parts of the diagrams have been truncated since the magnitude of the DC component (the preload) extends to the thousands. The clear difference in the distribution of the frequency components of the two signals reflects the variability of the BCG waveforms across people. The significant variation in amplitudes overall between the two signals will be discussed later. Given that the Nyquist frequency (half the sampling frequency) is 42.4 Hz, an alias of the EMI (50 Hz) appears at 34.8 Hz.

Figure 6. A sample load cell signal. The upper plot shows the raw signal, while the lower one shows its filtered counterpart. The filtration is carried out by high- and low-pass fircls1 filters with 0.5 and 20 Hz cutoff frequencies, respectively. The magenta lines’ intersections with the time axes highlight the delay caused by filtration; about 3.7 s.

Figure 7. A segment of a BCG signal being processed by the clustering algorithm, which is shorter than the usual case for convenience of display: (a) local maxima are identified; (b) two of the twenty-six feature vectors contained in the segment are highlighted, namely, vectors 9 and 20, with each one comprising 28 features, i.e., signal samples.

Figure 8. A dendrogram showing the complete hierarchical clustering of the sample signal presented in the previous figure. The magenta line represents the dissimilarity threshold (θ = 1.2) that is used to determine the final clustering.

Figure 9. Sample of the workings of the respiration rate calculation method. The upper plot depicts a respiration signal with red asterisks marking the detected local maxima, i.e., the inhalation peaks, while the corresponding calculated breath-to-breath respiration rate is shown in the lower plot.

Figure 10. ECG signal samples with QRS complexes marked with red asterisks: (a) the common form; (b) baseline fluctuation; (c) large EMI.

Figure 11. Visual evaluation diagram that contains both BCG and ECG signals, which have been rescaled and shifted for the convenience of display; their respective detected heartbeats and corresponding beat-to-beat heart rates; and the signal epochs’ markers that highlight the segment’s divisions, by which the signal was handled. This volunteer has been diagnosed with heart arrhythmia.

Figure 12. Two samples of raw load cell signals measured for a volunteer in a supine position with the sensor positioned (a) below the mattress (see Figure 2); (b) at the top side of the mattress (see Figure 3). Respiration and cardiac cycles are marked by circles and “x”, respectively.