Differentiating of Respiratory Noises Based on Higher Order Spectral Analysis

The paper describes a method for differentiating respiratory noises, such as normal (bronchial and vesicular) and pathological (crepitation, dry and wet rales, noise of pleural friction). The presence of specific pathological noises in the breathing sounds indicates certain diseases. The described method is based on the application of higher-order spectral analysis. The paper indicates the advantages of using such an analysis. Three-dimensional images of the cumulant function of the third order and bispectrum, as well as their contour images were obtained. Based on these images, characteristic features for different types of respiratory sounds were identified and described. These features can be informatively useful for the primary identification of certain pathologies of the human respiratory system.