Monitoring of breathing phases using a bioacoustic method in healthy awake subjects.

OBJECTIVE: To test the ability of a microphone recording system, located distal to the respiratory outflow tract, to track the timing of the inspiratory and expiratory phases of breathing in awake healthy subjects.
METHODS: Fifteen subjects participated. Breath sounds were recorded using a microphone embedded in a face frame in a fixed location in relation to the nostrils and mouth, while simultaneously recording respiratory movements by respiratory inductance plethysmography (RIP). Subjects were studied while supine and were instructed to breathe normally for 2 min: through their noses only (nasal breathing), during the first min, and through their mouths only (oral breathing) during the second min. Five subjects (test group) were chosen randomly to extract features from their acoustic data. Ten breaths (5 nasal and 5 oral breaths) from each subject were studied. Inspiratory and expiratory segments of breath sounds were determined and extracted from the acoustic data by comparing it to the RIP trace. Subsequently, the frequency spectrum of each phase was then determined. Spectral variables derived from the 5 test subjects were applied prospectively to detect breathing phases in the remaining 10 subjects (validation group).
RESULTS: Test group data showed that the mean of all inspiratory spectra peaked between 30 and 270 Hz, flattened between 300 and 1,100 Hz, and peaked again with a center frequency of 1,400 Hz. The expiratory spectra peaked between 30 and 180 Hz and its power dropped off exponentially after that. Accordingly, the bands ratio (BR) of frequency magnitudes between 500 and 2500 Hz to frequency magnitudes between 0 and 500 Hz was chosen as a feature to distinguish between breathing phases. BR for the mean inspiratory spectrum was 2.27 and for the mean expiratory spectrum was 0.15. The route of breathing did not affect the BR ratio within the same phase. When this BR was applied to 436 breathing phases in the validation group, 424 (97%) were correctly identified (Kappa = 0.96, P < 0.001) indicating strong agreement between the acoustic method and the RIP.
CONCLUSION: Frequency spectra of breathing sounds recorded from a face-frame, reliably identified the inspiratory and expiratory phases of breathing. This technique may have various applications for respiratory monitoring and analysis.