Noninvasive detection of coronary artery disease using parametric spectral analysis methods

The detection of coronary artery disease by noninvasive analysis of isolated diastolic heart sounds is considered. It is based on identifying features associated with turbulent blood flow in partially occluded coronary arteries. The application of two types of parametric spectral analysis-autoregressive methods and eigenvector methods-to identify the additional signal components is discussed. Results obtained with one eigenvector method, (the MUSIC method) for spectra obtained from an angioplasty patient and results obtained with the autoregressive model in a comparison study of ten diseased and five normal patients are presented and discussed.<
>