Wheeze Detection and Location using Spectro-temporal Analysis of Lung Sounds

Wheezes are abnormal lung sounds, which usually imply obstructive airways diseases. The objective of this study is to design an automatic wheeze detector for a wearable health monitoring system, which is able to locate the wheezes inside the respiratory cycle with high accuracy, and low computational complexity. We compute important features of wheezes, which we classify as temporal and spectral characteristics and employed to analyze recorded lung sounds including wheezes from patients with asthma. Time-frequency (TF) technique as well as wavelet packet decomposition (WPD) is used for this purpose. Experimental results verify the promising performance of described methods.