With the advent of the fast Fourier transform and the development of hybrid computing, the digital analysis of various biologic signals in the frequency domain has become a realistic objective. the application of spectrum analysis to biomedical problems is no longer restricted by either hardware or software limitations, and consequently one is now confronted with what is best described as a problem in data analysis. In this paper an application of spectrum analysis in phonocardiography is described. We propose that separate spectra be computed for the first and second heart sounds and suggest how the spectra can be used in both experimental and clinical investigations. The computational procedure is described, an example is presented, and a FORTRAN subroutine is provided that can be used to compute the spectrum estimates. The results presented here represent a first step in the data analytic process and provide a method of exposing and summarizing the “information” in the data. The method has potential application in both the diagnosis of cardiovascular disease and the monitoring of critically ill patients. To illustrate the latter point, we propose that the heart sound spectra could be used to construct an index of relative heart sound intensity that has potential application in monitoring the depth of anesthesia during surgery.
Read full abstract