Time-frequency analysis of wheezing sound to differentiate asthmatic and non-asthmatic condition

Abstract

In this paper, a new method to analyze wheezing sound to differentiate asthmatic and non-asthmatic condition is proposed. To achieve this, data acquisition was done on asthmatic and non-asthmatic patients. The data was then filtered by using the high pass-Butterworth filter to obtain a smooth signal. Segmentation of expiration phase emphasized wheezing signal characteristic of the total of 60 epochs. The next step was the selection of time-frequency distribution (TFD) which enabled the feature extraction of frequency, maximum energy, and average energy. Based on comparison done, Modified-B distribution exhibited the best time-frequency resolution for this application. Extracted wheezing features from the time-frequency distribution of asthmatic and non-asthmatic conditions were subsequently analyzed using statistical analysis of t-test. The result indicates that the frequency can be used to differentiate asthmatic and non-asthmatic condition. In conclusion, the Modified-B distribution can distinguish asthmatic and non-asthmatic condition, based on frequency extraction.