Estimation of respiratory flow rate by analysis of breath sounds at the external ear

Gary A Pressler, Purdue University


Respiratory monitoring during athletic activity or other clinical and occupational settings would benefit from a method that does not obstruct the airway. Previous studies have demonstrated the utility of breath sound measurements performed on the chest or neck to detect airflow. This dissertation investigates a novel location for breath sound measurements: the external ear. In all of the experiments, a miniature electret microphone, set within a disposable rubber earplug, was placed in each subject's external acoustic meatus. The first experiment demonstrated the passive transmission of sound from the oropharynx to the external ear in nineteen adults. Broadband noise was introduced to each subject's pharynx via a tubular mouthpiece and was recorded by both an accelerometer affixed to the subject's cheek and the ear microphone. Near-unity coherence estimates (> 0.9) were observed up to 800 Hz, indicating a low-frequency region of preferred transmission. The earplug provided at least 25 dB of acoustic isolation from environmental noise. In the second experiment, the ear microphone detected tidal and shallow breath sounds in twenty adults. A third experiment studied eleven adults to determine the relationship between respiratory flow rate and average sound power (300--600 Hz). Six different configurations were used: oral, nasal, and open (oral and nasal) breathing, each with and without a pneumotachograph. Sound spectra for a given subject and configuration maintained a characteristic shape but varied in amplitude with flow. The flow (F) to average sound power (P) relationship followed the form P= AFB. Values of B were typically between 3.5 and 5 and varied by subject, respiratory phase, and breathing route. The final experiment estimated flow rate from the breath sounds of ten subjects using the above relationship. Coefficients from one session calibrated sounds from the same subject two days and two weeks later. While the variance of individual flow estimates was relatively large for instantaneous flow predictions, average flow estimates typically had only 10% error for open (oral and nasal) breathing. Nasal breathing demonstrated relatively large variances between successive sessions. More complex mathematical relationships are likely needed for more accurate flow estimation from breath sounds measured at the ear.




Wodicka, Purdue University.

Subject Area

Biomedical research

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server