Flow And Volume Related AR-modelling Of Lung Sounds

Abstract

Usually, the analysis of phonopneumography (PPG, lung sound recordings) is implemented as an automated form of auscultation, in which spectral features of normal breath sounds as a function of time play an important role. In this paper, a new approach to PPG analysis is proposed. Ultimately, not the lime-dependence is important as such, but rather the dependence upon the phase within the respiratory cycle. Experimental results indicate that this dependence can be quantified by relating time-varying AR-coefficients of the PPG to simultaneously recorded flow and volume signals. This may allow to identify physiclogically meaningful quantities directly from recordings.