Computer analysis techniques for phonocardiogram diagnosis

An analysis procedure for small digital computers was developed to study heart sounds and heart murmurs in the frequency, time, and envelope domains. Signal processing techniques which can aid positive diagnosis are described and plots are presented to demonstrate their effectiveness. It is shown that heart sounds taken at the same respiration phase are nearly periodic in the systole. This property can be used to compute statistically reliable averaged envelograms, power spectra, and wavelets. The algorithms developed are applicable to may heart diseases, but the initial study is being conducted on pulmonic stenosis, aortic stenosis, and normal patients.