Acoustic impedance rhinometry (AIR): a technique for monitoring dynamic changes in nasal congestion

Many semi-quantitative measures of nasal patency have been used, each with its own advantages and disadvantages (e.g. visual inspection, nasal speculum, endoscopy, optical rhinometry, condensation on glass, or the 'hum-test'), but there are only three quantitative methods used commonly: rhinomanometry, acoustic rhinometry, and imaging (e.g. CAT scan). The first two are the most useful for studies of the dynamic (i.e. real-time) changes in nasal resistance, as described below.

Rhinomanometry requires measurement of air flow and pressure, with nasal airway resistance given by their ratio (pressure over flow; Kenyon and Pickett 2009). The required air flow through the nose is either generated by an external pump, or more commonly by normal air flow during the subject's tidal breathing. However, poor correlations with other methods and issues of repeatability with this technique have been cited (Hilberg 2002). This may partly be due to turbulence and tissue valving, where the relationship between flow and pressure (and therefore their ratio or resistance) is not linear for the higher flows and pressures generated during breathing (Hilberg 2002). In addition, rhinomanometry cannot make simultaneous measurements of left and right nasal passages (Hilberg 2002).

Acoustic rhinometry has emerged from techniques developed in geology and mining, where sound (clicks or wide-band noise) is sent into the nasal cavity, and cross-sectional area (not resistance per se) as a function of distance through the nose is calculated from the averaged reflected sound waveform by unravelling the multiple delayed echoes using the Ware-Aki algorithm (Hilberg 2002). The estimated cross-sectional area profile of the nasal passages has been shown to correlate moderately well with MRI scans of the nasal passages, but the technique has its limitations. It requires specialized clinical equipment, the minimum time between measurements is limited (due to averaging and echo duration), and is confounded by the variable link between cross-sectional area and resistance with different cross-section profiles (e.g. for the same cross-sectional area a circular duct has a lower resistance than a narrow duct).

Optical rhinometry (ORM) is a more recent technique which monitors the absorption of near-infrared light of different wavelengths as it passes laterally through the nose (as in pulse oximetry; Wüstenberg et al 2007). While elegant and convenient, ORM does not provide independent measurement of the left and right nasal passages, cannot distinguish between the upper and lower turbinates, and monitors equivalent tissue density rather than air-flow resistance.

Overall, there is a need for an inexpensive, convenient technique for the simultaneous bilateral measurement of nasal air-flow resistance, for research and/or clinical purposes. As a result, we have developed a simple objective method of quantifying nasal resistance by monitoring the small-signal acoustic input impedance of each nasal passage in real-time, using minimal and cost-effective equipment. The technique, which we refer to here as acoustic impedance rhinometry (AIR), is similar to the audiological measurement of ear drum compliance with acoustic tympanometry, and allows unilateral or bilateral recording with a time resolution of seconds, and the delivery of air-borne stimuli if required.

While we have previously used rhinomanometry to monitor nasal patency, using either active tidal breathing or an oscillating pump as a low-frequency airflow source, the method we describe here uses a small-signal high-frequency volume–velocity source: a small earphone delivering an acoustic probe tone through a long silicone tube (figure 1(a)). The acoustic probe tones were generated by a laptop through an inexpensive USB external sound card (Sabrent, USB-SND8), chosen because it could plug into any laptop and provided independent left and right microphone inputs. Sinusoidal probe tones from the two ear-buds were fed to the left and right nasal probes through 300 mm lengths of flexible silicone tubing of 1 mm internal diameter (figure 1(a)) so that each tube's high acoustic impedance ensured that the each ear-bud supplied an approximately fixed volume–velocity stimulus to each nosepiece. The sinusoidal pressure sensors were two small and inexpensive electret microphone elements, mounted in parallel with each ear-bud, and also separated from each nosepiece by a 300 mm length of tubing (1 mm internal diameter). Custom-written software developed in LabView 7.0 (National Instruments) allowed the monitoring of the raw pressure waveforms from the electret microphones, and their frequency spectra, specifically the amplitude of the probe tone which was proportional to the modulus of the nasal impedance. Sampling rate was 44.1 kHz with a sample epoch of 330 ms, with a running average over 10 epochs (3 s).

Detailed experimental notes were taken to allow annotation of impedance traces, and for later removal of any artefacts (e.g. swallowing, coughing). An additional 300 mm tube identical and in parallel to the two others was connected to each nose probe, and provided a low-frequency pressure release, normally to room air. This procedure minimized pressure transients during swallowing and other subject manoeuvres. The high acoustic impedance of these 'air delivery tubes' also ensured that they did not significantly shunt the nasal impedance under measurement. When attached to a small air pump, the third tube also allowed delivery of air-borne stimuli to each nostril.

Regarding nosepiece design, the series acoustic impedance of the acoustic probe within the nose was minimized by cutting all three tubes flush with the surface of the nose insert at the point where the tubes exited the insert into the nose. This ensured that the resistance registered as zero when the nosepiece was out of the nose in free air. Air leaks around the rigid cylindrical inserts were minimized by selecting their diameter to produce moderate tension around the nares on insertion, and tacky inert paper-glue ('Clear Gum', Bostik) was also applied around their edges.

Each subject's mouth was held open at a fixed gape using a 50 mm long section of corrugated plastic pipe of 25 mm diameter, with the tongue under the tube to minimize occlusion at the back of the throat (figure 1(b)). The tube's corrugations avoided buckling of the tube, prevented slippage, and provided a convenient rest place for the teeth. It was found that this open gape was sufficient to keep the component of series resistance contributed by the mouth and throat low and constant relative to nasal resistance. Subjects were also asked to swallow infrequently, and to keep their tongue low in their mouth to avoid glottal closure. Placing the tongue under the mouth-tube helped in this regard. Even so, there was always a component of the measured acoustic impedance that was due to the acoustic impedance through the back of the throat and out the mouth, and this was a common-mode impedance shared by the left and right nasal passages (figures 1(c) and (d)), which produced a small amount of acoustic cross-talk between the two channels (see later figure 3(a)). Because the nasal resistances and throat/mouth resistance were in series, the more occluded the nasal passage, the less significant was the contribution of the throat/mouth impedance.

In summary, an equivalent electrical circuit for our acoustic system is presented in figure 1(d) (mass reactances as inductors, acoustic resistance as electrical resistance, and acoustic compliances as capacitances). The physiological (nasal) component of the system including the left and right nasal resistances and shared throat/mouth resistance is shown above the horizontal dashed line, while the left and right nasal couplers, microphones, volume–velocity sources and stimulus delivery tubes are shown below the dashed line. Each ear-bud speaker (the acoustic equivalent of an alternating (ac) voltage source) is shown feeding into its high-impedance delivery tube, so that the ear-bud/tube combination approximates an acoustic volume–velocity source, analogous to an electrical ac current source. The air pump used to deliver chemical stimuli is the acoustic equivalent of a direct current (dc) source (dc_air), while each electret microphone is equivalent to a voltmeter, supplied through the high resistance of each acoustic probe tube. R_nose represents the variable resistances of the left and right nasal passages, respectively. The acoustic compliances of each nasal passage are represented by C_nose. The shared resistance at the back of the throat (R_throat) and the shunt resistance (R_mouth) and compliance (C_mouth) of the mouth are also shown. Closing the mouth during recording (e.g. while swallowing) dramatically increases R_mouth and confounds the recordings, but the mouth tube and open gape minimized the effects of this resistance.

The stability of the recordings is demonstrated in figure 2, where long-term (3000 s) recordings are shown for three normal subjects 30 min after nasal decongestion with oxymethazoline (500 µg mL⁻¹). Probe frequency was 433 Hz. The resting nasal resistances in the three subjects are expressed as a percentage of the calibration resistance, and varied very little around each individual's mean value over the 3000 s (means and standard deviation for the first 2500 s (before jugular occlusion in the first two subjects) were 6.39±0.40%, 9.77±0.85% and 6.68±0.27% relative to the calibration resistance. The transient rise in resistance at the end of figures 2(a) and (b) is due to a brief bilateral occlusion of the jugular veins to elevate cranial blood pressure, engorge nasal tissue, and elevate nasal resistance, to demonstrate that the measurement system was still fully functional. The small variability in these decongested noses was less than that observed when no decongestant was used (see other figures), suggesting normal dynamic changes in nasal resistance without decongestant, presumably under neural control. Overall, we experienced few problems with condensation within the measurement tubes with the 1 mm diameter silicone tubing (more problems were experienced with earlier trials with narrower polyethylene tubes with internal diameters of 0.7 mm). Because there could also be changes in measured resistance with changes in posture and blood pressure (as described later), subjects were asked to sit very still during recordings. Reproducibility within a single subject over days is demonstrated in figure 5 for one subject, where repeat recordings of nasal decongestion are shown. After decongestion, the resting nasal resistance on the two occasions was within a few percent.

Zoom In Zoom Out Reset image size

In this study we optimized the AIR technique to measure acoustic resistance (rather than compliance or mass-reactance), because resistance is important in nasal congestion, and is a good measure of the neurally-controlled dynamic changes of the nasal valve. We initially used a probe frequency of 217 Hz to avoid mains interference (multiples of 50 Hz in Australia), and because the frequency was high enough to allow the small ear-bud drivers to produce sufficient sound level for an adequate signal-to-noise ratio (SNR) at the nasal entrance. As the sensitivity of our equipment improved during the development process, however, a higher drive frequency of 433 Hz was used to improve the SNR in the presence of an unwanted acoustic noise from the air pumps used to deliver a through-flow of air-borne stimulants.

Because the two signal channels of our bilateral measurement system monitored the magnitude of the pressure signal, without routinely recording its phase, we mostly measured only the modulus of nasal acoustic input impedance. That is, we did not routinely discriminate between acoustic resistance, acoustic compliance or mass-reactance. Nevertheless, even at 433 Hz the acoustic impedance was dominated by resistance, with the phase of all measurements within 10° of the phase of the response at the opening of the resistive calibration tube (see later figure 5). These phase-sensitive recordings were done by substituting a fixed phase reference signal (the 433 Hz drive sinusoid) for the microphone signal from one nostril, producing a single-channel phase-sensitive system.

With this arrangement, our 433 Hz probe sound generated a sinusoidal pressure at the entrance of our dummy nose (a resistive calibration tube, see later) that was 104 dB SPL under standard conditions. In a normally functioning nasal passage, with a resistance 10–20% of our standard calibration tube, the sound level within the sealed nasal entrance was between 84 to 90 dB SPL, and the probe tone could just be heard by uncongested subjects as a quiet pure tone (with no higher harmonics present).

To allow checking of the throat/mouth impedance and to improve separation of the contributions from the left and right nasal impedances (i.e. reduce left/right cross-talk), we trialled using different probe frequencies for left and right nasal passage, and also alternated presentation of the left and right probe-tones. The simpler solution that was eventually adopted (requiring less software and producing a simpler data output) was the manual pinching of each sound delivery tube in succession, so that the consequent decrease in signal from the contralateral nasal passage indicated the magnitude of 'cross-talk' between left and right measurements at that time. This is illustrated in figure 3(a).

Zoom In Zoom Out Reset image size

The system could be calibrated conveniently with a calibration tube, made from the barrel of a 1 ml syringe (Terumo) with its tip removed at the 0 ml mark (figure 3(b)). This standard resistance was readily available, easily replicated, and easily understood. Its (laminar flow) resistance could be calculated using Pouiselle's law to be approximately 200 kNs m⁻⁵ (confirmed by direct measurement with air flow from a gas bottle, data not shown). This calibration resistance was coupled to the nose probe with a silicone test-tube stopper (which sometimes needed petroleum jelly to ensure an airtight seal). Software-scaling could be adjusted so that the calibration tube corresponded to a 100% reading, and in the remainder of this report nasal impedances are expressed as a percentage of the impedance of this standard calibration tube. Overall, we found that this calibration procedure was only needed occasionally (e.g. to check for leaks in the delivery/measurement tubes and nose probe), although calibration was always confirmed in every measurement. Figure 3(b) shows an example of the calibration procedure. The slow rise and fall in measured resistance was due to the moving point average of the measurement system, and not to any significant delay in the resistance change (which was instantaneous).

The stability of the measurement technique was evident in biological measurements of nasal passages in non-hypersensitive (i.e. non-allergic) subjects, and in nasal passages rendered inert and uncongested by pre-treatment with pharmacological decongestants (figures 2 and 4). Removal and reinsertion of the nasal inserts did not cause significant changes in the measured resistance and impedance phase (see figures 2 and 5). For protracted measurements, swallow artefacts were evident, because swallowing caused transient artefactual changes in the measured resistance due to microphone distortion during the low-frequency transients (figure 4(a)). However, measurements remained stable before and after the swallow artefact. Detailed experimental notes were kept during measurements, recording the time of each subject swallow, to allow removal of swallow artefacts when they occurred. As a result, the fluctuations in nasal resistance we have observed were not due to measurement instability or artefacts, but represented true biological variations in nasal resistance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Pharmacological decongestion of the nose proved to be the simplest and most effective means of producing large changes in nasal resistance to demonstrate the efficacy of the AIR technique. The active ingredients in over-the-counter decongestants are sympathomimetics (e.g. phenylephrine and oxymetazoline) that reduce blood flow to nasal capacitance vessels by directly stimulating α-adrenergic receptors in the afferent arteriolar smooth muscle (Proctor and Adams 1976). The effect of phenylephrine or oxymetazoline on nasal resistance is illustrated in figures 5 and 6, with data from one typical subject. All decongestant recordings were made at least two days apart, because of the long-lasting effects of the decongestants on the nasal turbinates (including rebound effects).

Zoom In Zoom Out Reset image size

Figure 5 shows two separate trials of nasal decongestion in one subject. After a demonstration of the stability with removal and reinsertion of the nasal probe, decongestant (phenylephrine) was applied bilaterally, during which unilateral (left side) measurements of nasal impedance were made (amplitude and phase, measured relative to the calibration resistance tube). In both trials, nasal resistance dropped by 20 to 30% over 375 s, to plateau at a stable baseline of approximately 20% calibration resistance (a resistance of approximately 40 kNs m⁻⁵; figures 5(a) and (b), upper panels). During decongestion, the phase of the measured impedance remained within 10° of the impedance phase of the resistive calibration tube (figures 5(a) and (b), lower panels), indicating that the impedance of the congested and uncongested nasal passage is dominated by resistance. Removal and reinsertion of the nasal probes caused impedance phase rolls of less than 5°, also indicating that there was little change in impedance with small trial-to-trial variations of the depth to which the nasal probes are inserted into the nares.

Figure 6 demonstrates the effect of unilateral (right side) decongestant application on the nasal passages, showing a decrease in resistance bilaterally, but with a slightly faster decrease in the nasal passage exposed directly to the decongestant. It is not clear if the bilateral drop in resistance is due to a spread of decongestant across the midline through blood flow, or more likely due to bilateral neural control of blood flow.

As another demonstration of the utility of the AIR technique, the effect of posture on nasal resistance was investigated, on the basis that changes in hydrostatic pressure in the nasal blood vessels should cause measurable changes in nasal patency by changing nasal tissue volume (assuming incomplete compensation by any neural reflexes). As illustrated in figure 7, fast increases in nasal resistance were seen when a subject moved from standing to supine, consistent with passive filling of the capacitance vessels with an increase in cranial blood pressure on lying down. Others have previously noted changes in nasal patency with posture, but were limited to monitoring changes over a much longer time course, largely because their measurement procedures were slower than the AIR technique (Kase et al 1994). A similar change with altered cranial blood pressure was also evident with brief jugular blockade (figures 2(a) and (b)).

Zoom In Zoom Out Reset image size

The main impetus in our development of the AIR technique was our interest in environmental stimuli that modulate nasal resistance, particularly in rhinitis. In particular, we had hypothesized that afferent input (temperature sensors, nociceptors, and other afferents) in the nasal cavity may integrate to alter blood flow to the conchae and alter nasal resistance. As an example of the sensitivity of the nasal tissue to direct action of air-borne stimuli, ammonia was used to produce rapid nasal congestion (figure 8). The stimulus-delivery air pump's inlet hose was held briefly above a solution of ammonia (window cleaner). As shown in figure 8(a), when brief (30 s) pulses of ammonia vapour were delivered simultaneously to both nasal passages via the stimulus-delivery tube, nasal resistance increased bilaterally over approximately 125 s by 10% on the right, and 15% on the left. A similar time course was observed with unilateral ammonia application, as shown in figure 8(b). Left and right nasal resistance was measured simultaneously while ammonia vapour was applied unilaterally via the stimulus-delivery air pump, as indicated. There was a clear, reproducible increase in resistance in the nasal passage to which the ammonia stimulus was applied, approximately 100 to 125 s following the stimulus.

Zoom In Zoom Out Reset image size