A microcomputer based lung sounds analysis

The use of a microcomputer in lung sound-analysis is described. The system was used experimentally in order to evaluate automated auscultation as a mean for improving the sensitivity of pulmonary health mass screening. The sound signals from four custom-made piezoelectric transducers, affixed at specific locations on the chest wall, and the breathing flow signal produced by a pneumotachograph were amplified, filtered and digitized simultaneously at 4000 Hz per channel for 512 ms. The acoustic data were transformed to the frequency domain to enable the calculation of the power spectra. Those were averaged over successive runs and displayed as log power vs. frequency. The operator could assess the convergence of the spectral pattern using the on-line graphics and calculated parameters, and store the data once the noise level had reached a preset level. This procedure was repeated during expiration, inspiration and on breath arrest. The results of the off-line analysis of the lung sounds, combined with pulmonary function tests and a questionnaire, were used to identify lung pathology.