Ballistocardiogram (BCG)

Ballistocardiogram measures the movements of the body which are caused by the hifts in the center of  mass of the blood and to a lesser extent of the heart [1]. The first BCG recording was  published  in 1877 but the modern ballistocardiography in considered to have begun  in 1936 by Isaac Starr when he built a new type of bed BCG measurement  device. Since then various type of bed and chair type BCG measurement devices have been constructed. A lot of scientific work was done around BCG in 1940-1975 but some of it was performed without proper knowledge of physiology and physics which gave a somewhat questionable reputation to BCG . As the recording and particularly the understanding of the electrocardiography, ECG improved and ECG had a better specificity, the use of BCG decreased dramaticalyy in the 1970'ies. BCG and related studies are nowadays practiced in the research community surrounding the Cardiovascular System Dynamics Society (CSDS). Improved sensor technology has attracted more interest in BCG in recent years.

BCG waveform
In the early days the measurement devices measured so-called displacement BCG and sometimes velocity BCG, its first derivative. Nowadays the equipment usually measure the acceleration BCG which is the second derivative of displacement BCG. This is so, because the modern sensors for BCG measure force which is mass times acceleration. The theoretical BCG waveform can seen in Figure 1 and its prctical  counterpart in Figure 2. Smith recommends that the filters in the measurement process should have a pass band of 0.3-50 Hz [1}, but practically all power in the signal is below 20 Hz and at 0.5 Hz one still records quite a lot of respiratory components.

Theoretical BCG waveform
Figure 1. A theoretical BCG waveform and its components. The extrema of the BCG waveform are denoted with letters F, G, H, I, J, K, L, M and N. The waves are divided in three groups: pre-ejection (FGH), ejection (IJK) and diastolic (LMN). The R-spikeof the ECG can give timing reference.
Practical BCG waveform and related signals
Figure 2. A practical BCG waveform and some othe signals for comparison. From top to bottom: Impedance cardiography, electrocardiography, BCG measured from a flat sensor in a seat of a chair and a BCG measured from the back of the chair. In this measurement the subject is breathing normally and this introduces the slow fluctuation to the baseline of particularly the back BCG signal.
Interpretation of BCG
Several ways to analyse the BCG waveform have been presented in the literature. One of the most well-known methods to analyse the wvaform is by  Isaac Starr in 1966 [2]. Starr classifies the BCG into four groups, I-IV based on a recording of one complete respiratory cycle. Group I consists of  normal subjects with a steady train of basic BCG waveform patterns. In group II a minority of the BCG waves are abnormal. The majority of the BCG waves are abnormal in group III. In group IV, the BCG waveforms are so disorganized that the positions of the major BCG waveform components cannot be identified properly. Experts can see the signs of certain pathologies by simply examining the BCG recordings visually. E.g. idiopathic hypertrophic subaortic stenosis [3-6]  produces visually distinguishable patterns. BCG offers also good perspectives for application in preventive medicine, particularly in the early detection of coronary heart disease [7].

BCG can also be analysed quantitatively by measuring the amplitudes, time intervals, slopes etc. of the BCG waveforms. When interpreting the results the fact that the BCG amplitudes decrease with age must be taken into account.  Perhaps the most studied amplitude measure is the I-J amplitude. The later amplitudes are less robust measures as their amplitude is smaller and thus more subject to background and other variations. The I and IJ amplitudes have been mentioned to be useful for detecting subclinical and early abnormalities in large populations, in testing effects of drugs and other therapy, in evaluating certain diseases such as aortic valvular disease, in detecting the presence of coronary artery disease, and in predicting life expectancy [1]. The time intervals of the waveforms are sometimes studied, as well. These can be measured, e.g. with reference to the R-spike of the ECG.
[1] Smith N.T., Ballistocardiography in Weissler A.M. (Ed.),  Noninvasive cardiology. Grune & Stratto, New York, USA, 1974.
[2] Starr I., Proc. First Congress Ballistocardiography and Cardiovascular Dynamics, Amsterdam, Karger; New York, USA, pp. 7-20, 1966.
[3] Jackson D.H., Eddleman E.E. Jr., Bancroft W.H. Jr., et al., Ballistocardiographic and angiographic correlative study in idiopathic hypertrophic subaortic stenosis. Bibl. Cardiol., Vol. 27, pp. 14-20, 1971.
[4] Deuchar D.C., Some further observations on the ballistocardiogram in aortic valve disease. Proc. 3rd European Symp. Ballistocardiography, Rochet, Brussels, pp. 225-233, 1962.
[5] Elsbach H., Rodrico F.A., Westerhof N., The ultralow frequency ballistocardiogram in muscular subaortic stenosis. In Knoop A.A. (Ed.), Ballistocardiography and Cardiovascular Dyanmics, Basel Karger, pp. 26-30, 1966.
[6] Elsbach H., Rodrico F.A., Verbiest E., Further observations on the ballistocardiogram in hypertrophic cardiomyopathy. Bibl. Cardiol., Vol. 21, pp. 125-129, 1968.
[7] Goedhard W.J. Ballistocardiography: past, present and future. Bibl. Cardiol. Vol. 37, pp. 27-45, 1979.
Alpo Värri 26.8.2008