Digital processing techniques of breath sounds for objective assistance of asthma diagnosis

We describe a method based on the acoustical analysis of breath sound for helping the diagnosis of lung diseases. The signal is recorded at the trachea simultaneously with the flow rate at several values of this one. The signal is digitized at the rate of 5120 Hz, FFTs are taken on 1024 sample blocks over a duration of about 30 seconds. We plot versus time the mean frequency and the mean amplitude of the spectrum, the flow rate and it slope. We compute also the mean frequency spectrum. We obtain two kinds of results. First the variations of the mean frequency in function of the flow rate give us informations on the breath sound formation. Second the differences in mean frequency spectra between asthmatic and normal subjects give us a good discrimination of the two cases.