Session S61.4

Respiration Analysis of the Sternal Ballistocardiograph Signal

K Tavakolian*, A Vaseghi, B Kaminska, H Kennedy-Symonds

Simon Fraser University
Burnaby, BC, Canada

It is shown in many different research works in the past that there is a considerable respiration effect on the Ballistocardiograph (BCG) signal. The pre-processing approach that was taken by previous research works was simply the removal of respiration signal by high-pass filtering of respiration from the BCG signal and in some other research works breath holding was advised during the BCG recording to remove the respiration effect while holding the breath changes the normal morphology of the BCG signal.
The novelty of our research is that, rather than trying to remove the respiration effect from the BCG signal by either filtering or breath holding, we utilize the respiration information to improve the averaging and thus the analysis and interpretation of the BCG signal in diagnosis of cardiac malfunctions. This methodology is based on our investigation that BCG cycles corresponding to the inspiration and expiration phases of respiration cycle are different in morphology.
The Sternal BCG together with respiration and ECG signals were acquired from 45 subjects in Burnaby General Hospital before and after exercise on the treadmill. In order to do a statistical analysis and as a measure of similarity, all the inspiration cycles were cross correlated, and the average of all their correlation coefficients, compared to one another, was considered as the measure of similarity of the inspiration cycles for that individual subject. The same procedure was repeated for expiration cycles and for all the forty five subjects to produce the table of results from which it was noticed that expiration cycles are 0.1272 more in the average similarity index, compared to inspiration cycles for pre exercise signal. A t-test was used to find out whether mean difference in a response variable, consisting the difference between expiration and inspiration values, was equal to zero. The analysis showed a significant statistical difference (p-value<0.01) between the inspiration and expiration averages. Processing the data, it was proven statistically that BCG cycles corresponding to expiration phase of respiration are more closely related to each other when compared to BCG cycles corresponding to inspiration and therefore, expiration cycles are better candidates to be selected for the calculation of the averaged BCG signal. This can be attributed to the fact that expiration is a passive process in which muscles relax and the thorax moves more smoothly, while inspiration is an active process in which the muscles contract and add more movement artifacts to the BCG signal which is placed on the sternum.
The focus of this research work was the introduction of a novel averaging method to the general BCG processing methodology which considers the respiration, to calculate the BCG average. This new BCG template can be shown to the user as the output of the BCG instrument or be used for further pattern recognition and interpretation of BCG signal.

(Abstract Control Number: 233)