Heart-Sound Processing by Average and Variance Calculation - Physiologic Basic and Clinical Implications

A new statistical method for heart-sound processing was developed and tested on normal subjects and on patients suffering from various cardiac pathologies. The method is effective in decreasing noise and in separating heart sounds from murmurs, as well as in deriving new physiological parameters. The theory is based on the assumption that heart sounds can be classified into deterministic and nondeterministic sounds. The processing results in a very significant attenuation of strong murmurs, while the deterministic events, such as SI-S4, are only slightly affected. The method includes dividing the heart-sound signal into a set of repetitive signals (ensemble) according to the trigger selected to be the peak of the ECG R-wave. The variability of the time elapsed from the trigger to the evoked sound is defined as the jitter. The average and variance functions are calculated from the ensemble. Calculation of the heartsound jitter from the average and variance functions shows a jitter of 5.5 ms ±2.6 ms for S 1 , and 8.2 ms ±3.3 ms for S 2 . The jitter, which is an objective parameter of the trigger-response linkage, can be used experimentally to clarify some of the cardiac electromechanical mechanisms, and it may have diagnostic value.