A signal processing approach for the diagnosis of asthma from cough sounds.

Asthma is not very easy to be correctly diagnosed by physicians where asthma is considered a common chronic inflammatory disease. Distinguishing of cough sound can be used to diagnose asthma. The use of signal processing techniques of cough sound for detecting asthma will be addressed in this paper to help the diagnosis of asthma by physicians. Since cough sounds are non-stationary and are stochastic signals inherently, time-frequency transform techniques are used to deal with such signals. Time-frequency analyses are performed to show in a comprehensive approach the characteristics of the cough sound signal. Time-frequency analysis techniques, specifically Wigner distribution in addition to wavelet transform to analyse cough signals, are used in this paper. The features extracted from the time-frequency domain of the cough sound are used as inputs to the asthma and non-asthma classifier. The results of the proposed algorithm are competitive to the best existing algorithms in the literature.