Acoustic reflectometry imaging of the airway

Computerised lung sound monitoring to assess effectiveness of chest physiotherapy and secretion removal: a feasibility study

OBJECTIVES

To explore the feasibility of computerised lung sound monitoring to evaluate secretion removal in intubated and mechanically ventilated adult patients.

DESIGN

Before and after observational investigation.

SETTING

Intensive care unit.

PARTICIPANTS

Fifteen intubated and mechanically ventilated adult patients receiving chest physiotherapy.

INTERVENTIONS

Chest physiotherapy included combinations of standard closed airway suctioning, saline lavage, postural drainage, chest wall vibrations, manual-assisted cough and/or lung hyperinflation, dependent upon clinical indications.

MAIN OUTCOME MEASURES

Lung sound amplitude at peak inspiration was assessed using computerised lung sound monitoring. Measurements were performed immediately before and after chest physiotherapy. Data are reported as mean [standard deviation (SD)], mean difference and 95% confidence intervals (CI). Significance testing was not performed due to the small sample size and the exploratory nature of the study.

RESULTS

Fifteen patients were included in the study [11 males, four females, mean age 65 (SD 14) years]. The mean total lung sound amplitude at peak inspiration decreased two-fold from 38 (SD 59) units before treatment to 17 (SD 19) units after treatment (mean difference 22, 95% CI of difference -3 to 46). The mean total lung sound amplitude from the lungs of patients with a large amount of secretions (n=9) was over four times 'louder' than the lungs of patients with a moderate or small amount of secretions (n=6) [56 (SD 72) units vs 12 (13) units, respectively; mean difference -44, 95% CI of difference -100 to 11]. The mean total lung sound amplitude decreased in the group of 'loud' right and left lungs (n=15) from 37 (SD 36) units before treatment to 15 (SD 13) units after treatment (mean difference 22, 95% CI of difference 6 to 38).

CONCLUSION

Computerised lung sound monitoring in this small group of patients demonstrated a two-fold decrease in lung sound amplitude following chest physiotherapy. Subgroup analysis also demonstrated decreasing trends in lung sound amplitude in the group of 'loud' lungs following chest physiotherapy. Due to the small sample size and large SDs with high variability in the lung sound amplitude measurements, significance testing was not reported. Further investigation is needed in a larger sample of patients with more accurate measurement of sputum wet weight in order to distinguish between secretion-related effects and changes due to other factors such as airflow rate and pattern.

Fitting the frequency response in acoustic reflectometry for an in vitro model of a human upper airway

Acoustic reflectometry has been used earlier to estimate the bore profile of a human upper airway and endotracheal tubes using thousands of acoustic pulses. The axial resolution of the profile computed depends on generating a broad frequency response, filtering undesired frequencies and compensating for wave propagation losses. The work presented here shows that acoustic reflectometry may be performed by using few Gaussian-sinusoidal acoustic waves, equally spaced along the bandwidth, and the Ware-Aki algorithm. In this way, the frequency content is controlled, the computational cost is reduced, and the axial resolution is improved. This entails fitting the frequency impulse response, compensating for propagation losses in the amplitude response, and analyzing the phase response. The methodology was validated in the laboratory with an in vitro model of a human upper airway, coupled to an acoustic reflectometer. Only one hundred acoustic waves, with a frequency step of 100 Hz, were needed to reconstruct a profile with an axial resolution within an acceptable margin of error.