Detection of COPD's diagnostic signs based on polyspectral lung sounds analysis of respiratory phases

This study examines the method for determining the specific auscultatory diagnostic signs in patients with chronic obstructive pulmonary disease (COPD). The main idea of the method is based on the finding of these features in separate respiratory phases of lung sound. For this purpose the original algorithm which divides the sound signal to inspiration and expiration phases was developed. Each phase was analyzed separately using polyspectral analysis methods. Diagnostic features were determined on the basis of the construction of three-dimensional bicoherence function and calculating the bicoherence and skewness coefficients. The set of all calculated characteristics made it possible to evaluate each respiratory phase. This allows to conclude about the presence or absence of auscultatory artifacts as pathology indicators.