Comparative analysis of phase difference estimation methods quantifying asynchronies between compartmental chest wall volume signals

Abstract

Asynchronous breathing movements may be observed in the presence of pulmonary disease, such as chronic obstructive pulmonary disease (COPD). This study was undertaken in an attempt to propose a reliable methodology to quantify this asynchrony. Five methods for estimating phase differences between two signals, based on the phase angle of the Fourier Transform (PhDFT), paradoxical motion (PhDPM), the Lissajous figure (PhDLF), maximal linear correlation (PhDP) and least-squares filtering (PhDLS), were compared. Frequency-modulated signals, simulating compartmental chest wall volumes, were used to evaluate the methods. Breathing asynchrony was quantified in two ways; by estimating (a) a single PhD value for the entire recording and (b) time-varying PhDs, representing non-stationarities of human breathing. PhDPM and PhDLF had the lowest average errors (4%), and PhDLS had a slightly higher error. PhDFT had zero error when estimating a single PhD value but a considerable error when estimating time-varying PhDs. PhDP presented the highest errors in all cases. An application of this methodology is proposed in real compartmental chest wall volume signals of normal and COPD subjects. Preliminary results indicate that the methodology is promising in quantifying differences in asynchronous breathing between thoracic volumes of COPD patients and healthy controls. ©2009 IEEE.