J Appl Physiol 98: 2235-2241, 2005.
First published February 3, 2005; doi:10.1152/japplphysiol.00683.2004
8750-7587/05 $8.00
Velocity and attenuation of sound in the isolated fetal lung as it is expanded with air
Philip J. Berger,1
Elizabeth M. Skuza,1
C. Andrew Ramsden,2 and
Malcolm H. Wilkinson1
1Ritchie Centre for Baby Health Research, Monash Institute of Reproduction and Development, Monash University, and 2Department of Newborn Services, Monash Medical Centre, Clayton, Australia
Submitted 1 July 2004
; accepted in final form 26 January 2005
 |
ABSTRACT
|
|---|
We measured the velocity and attenuation of audible sound in the isolated lung of the near-term fetal sheep to test the hypothesis that the acoustic properties of the lung provide a measure of the volume of gas it contains. We introduced pseudorandom noise (bandwidth 70 Hz–7 kHz) to one side of the lung and recorded the noise transmitted to the surface immediately opposite, starting with the lung containing only fetal lung liquid and making measurements after stepwise inflation with air until a leak developed. The velocity of sound in the lung fell rapidly from 187 ± 28.2 to 87 ± 3.7 m/s as lung density fell from 0.93 ± 0.01 to 0.75 ± 0.01 g/ml (lung density = lung weight/gas volume plus lung tissue volume). For technical reasons, no estimate of velocity could be made before the first air injection. Thereafter, as lung density fell to 0.35 ± 0.01 g/ml, there was a further decline in velocity to 69.6 ± 4.6 m/s. High-frequency sound was attenuated as lung density decreased from 1.0 to 0.5 g/ml, with little change thereafter down to a density of 0.35 ± 0.01 g/ml. We conclude that both the velocity of audible sound through the lung and the degree to which high-frequency sound is attenuated in the lung provide information on the degree of inflation of the isolated fetal lung, particularly at high lung densities. If studies of sound transmission through the lung in the intact organism were to confirm these findings, the acoustic properties of the lung could provide a means for monitoring lung aeration during mechanical ventilation of newborn infants.
fetal lamb; lung expansion; lung density; velocity of sound; sound attenuation
IN A STUDY THAT HAS PROVEN highly influential, Rice (16) estimated the velocity of sound in the parenchyma of the excised horse lung by measuring the time it took for a sound impulse to travel between two points on the pleural surface of the lung. He reported that inflation of the lung with air causes the velocity of sound passing through the parenchyma to increase from 
30 m/s at a low gas volume to 60 m/s with the lung maximally inflated. He further demonstrated that, when the lung was inflated with helium or sulfur hexafluoride, for which the free-field sound speeds are 970 and 140 m/s, respectively, there was little change in the measured velocity, from which he concluded that sound introduced at the pleural surface travels as longitudinal waves through the bulk of the parenchyma and not along the airways. His analysis indicated that average lung density and gas stiffness (compliance; approximately constant for diatomic gases) are the important determinants of sound velocity, and, as the average lung density depends on the relative proportion of the gas and tissue phases, measurement of velocity could be of potential use in determining lung condition.
A number of studies have confirmed Rice’s findings in lungs inflated in the range from residual volume to total lung capacity (7, 8, 10, 11, 17). The upper and lower limits of this range are of particular clinical significance, because compelling evidence shows that injury to the lungs during mechanical ventilation can be attributed to the effects of both overinflation, or volutrauma (5), and underinflation, or atelectrauma (18), in which there is repetitive opening and closing of underexpanded and unstable air spaces. The latter is especially relevant in preterm infants with respiratory distress syndrome, a condition characterized by surfactant deficiency, poor aeration, and lung collapse (3). While studies of the acoustic properties of the lung have been motivated by a desire to find a method for monitoring lung inflation, to date there has been little effort directed at assessing the utility of the approach in the poorly inflated lung, even though the theory outlined by Rice (16) suggests that velocity of sound through the lung should change substantially in this inflation zone.
In this study, we have used the isolated lung of the late-gestation fetal sheep as the model for assessing whether the acoustic properties of the lung reflect the volume of gas that it contains. By starting from a point at which the lung is completely free of gas, our approach has allowed the velocity and attenuation of audible sound in the isolated lung to be characterized as it is inflated from the liquid-filled state up to total lung capacity with air.
|
MATERIALS AND METHODS
|
|---|
All surgical and experimental procedures conformed with the guidelines established by the National Health and Medical Research Council of Australia and had the approval of the Standing Committee in Ethics in Animal Experimentation of Monash University.
Anesthesia was induced in near-term Border-Leicester ewes with intravenous propofol (5 mg/kg; Zeneca). The ewe was then intubated, and anesthesia was maintained with 66% N2O/1–2% halothane/32–33% O2. The maternal abdomen was opened in the midline, and the fetal head and neck were delivered through a uterine incision. After incising the neck, a liquid-filled tube, closed at one end with a stopcock, was advanced into the fetal trachea in the direction of the lung, with care being taken to prevent air entry. The tube was tied in place, and the ewe and fetus were killed with an overdose of anesthetic delivered to the maternal jugular vein (150 mg/kg Lethabarb, Virbac, Peakhurst, NSW, Australia). The lung and trachea were carefully dissected from the fetus, making sure that the lung surface was not damaged. The lung was weighed accurately using a Mettler balance.
Experimental Procedure
Each of the eight liquid-filled lungs that we studied, which weighed 177.0 ± 11.5 g when first dissected from the fetus, was positioned on a metal stage designed to allow sound to be introduced at its diaphragmatic surface (Fig. 1). At the center of the stage, there was a circular orifice 6 mm in diameter below which was mounted an electromagnetic sound transducer (CS-2218, Jaycar Electronics, Melbourne, Victoria, Australia) capable of producing broadband noise. The orifice dimensions were chosen so as to be 0.1 wavelengths in diameter at the highest frequency used (7 kHz). This criterion was adopted so that the orifice would act as a point source of acoustic radiation launching spherical sound waves into the fetal lung. An omnidirectional electret microphone (AM-4008, Jaycar Electronics) was attached via a double-sided adhesive electrode collar (32 mm outer diameter, 11 mm inner diameter; DM Davis) to the free surface of the lung in a position directly above the orifice. The microphone was coupled to the pleural surface using a small air chamber with a depth of 2 mm and a diameter of 9 mm. The distance between the diaphragmatic surface of the lung (taken as the surface of the sound stage) and the microphone was measured accurately with the aid of a digital Vernier scale attached to the side of the stage. This distance measurement was used when determining the acoustic attenuation of the lung tissue to correct for the reduction in sound intensity with distance expected in a spherically expanding acoustic wave (see below). A syringe was attached to one port of the stopcock on the tracheal tube, and a pressure transducer was connected to the other to measure pressure in the lung.
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 1. Schematic of the experimental setup, showing the lung positioned on a metal plate immediately above an electromagnetic sound driver, which injected pseudorandom noise into the lung through a 6-mm-diameter orifice. An electret microphone was positioned on the lung surface directly above the orifice, and the same microphone could be positioned directly over the orifice in the metal plate at the start of the experiment to determine precisely the intensity and timing of the input noise to the lung. The acoustic path length between the orifice and the microphone was measured with a vernier calliper. The lung was inflated using a graduated syringe attached to the trachea, and the intrapulmonary pressure was measured with a transducer attached at the input to the trachea. h, Height.
|
|
The first measurement of the acoustic properties of the lung was made with fetal lung liquid occupying the potential air space of the lung, i.e., there was no air present in the lung. To do so, we used CLIO software (AUDIOMATICA SRL, Firenze, Italy) to generate a pseudorandom voltage to actuate the electromagnetic sound transducer, producing a pseudorandom noise signal that passed through the lung to the microphone on the opposite side. The microphone output was amplified and filtered using a fourth-order linear phase band-pass filter with a lower cutoff frequency of 100 Hz and an upper cutoff frequency of 20 kHz. The strong filtering applied <100 Hz was used to limit potential interference from low-frequency room noise and the upper frequency limit to satisfy the Nyquist criterion for sampling at 48 kHz. In all experiments, the sound pressure level input to the lung produced by the electromagnetic sound transducer was set at 120 dB. The transfer function in the frequency domain was measured using the maximum length sequence method in which the impulse response of the lung is first measured by cross-correlating the pseudorandom sequence with the microphone output. Fast Fourier transformation of the impulse response then provides the magnitude and phase of the transfer function. The maximum length sequence was 16,384 samples long, and the sample rate was 48 kHz, giving a sequence period of 0.34 s. For each measurement, this sequence was repeated nine times to ensure a high signal-to-noise ratio (SNR). As a prelude to each experiment, the noise baseline was determined using the method just described but without the acoustic transducer activated to establish the SNR. For all measurements referred to in this paper, the ratio of desired signal to baseline noise was >20 dB, equivalent to a coherence of better than 91% (21). In addition, in each study, repeat measurements of the transfer function were made to confirm adequate SNR and showed <1-dB variation between each measurement across the frequency range of 100 Hz to 7 kHz.
After making the first measurement, we withdrew as much liquid as possible from each lung (25.6 ± 4.5 ml) and again measured the acoustic properties of the lung and the distance between the sound stage and microphone. Further measurements of these three variables, as well as intrapulmonary pressure, were then made after injection of known volumes of air into the lung, starting with 20 ml and progressing to 40, 60, 80, 100, 120, 180, and 240 ml. In some preparations, the lung developed a leak at higher volumes; acoustic data at this volume were excluded from the analysis.
Data Analysis
Acoustic properties of the lung.
As previously described, by cross-correlating the input pseudorandom noise signal and the recorded noise at the microphone, using CLIO software, we obtained the impulse response of the lung in the time domain at each level of lung inflation.
Lung density.
We calculated the density in the lung at each acoustic measurement, according to the relationship: lung density = lung weight/(volume of air in the lung + volume of lung tissue and liquid). Lung weight was taken to be the weight after dissection from the fetus, less the weight of liquid removed between the first and second measurements. To calculate gas volume at each step of the protocol, we used Boyle’s law to correct the volume of gas that had been injected into the lung based on the measured pressure in the lung. The volume of the lung tissue and liquid (in ml) at the start of the experiment was obtained from the weight of the lung immediately after dissection, assuming a tissue density of 1.0 g/ml.
Velocity of sound through the lung.
To calculate sound velocity at each inflation volume, we divided the distance between the sound stage and the microphone on the surface of the lung by the time interval between pseudorandom noise first reaching the lung surface and the start of the impulse response; the former value was established at the start of the experiment by placing a microphone directly on the sound stage surface and measuring the delay between start of the pseudorandom noise sequence in CLIO and its arrival at the microphone. After calculating velocity in this way, it was clear that our values were substantially greater than those reported by Rice (16) in the isolated horse lung. In view of the likelihood that sound transmission in the lung is dispersive, with high frequencies traveling faster than low (9, 17), we also used (unpicked) phase data obtained from the fast Fourier transformation of the impulse response to calculate the phase delay and hence the phase velocity over the frequency range of 500 Hz to 5 kHz. To do so, we used the phase data for the solid lung (before introducing air) as the reference for data for all inflation volumes; that is, we determined phase angle at each frequency and inflation level and then subtracted the phase angle at that frequency determined for the solid lung. This procedure allows for any phase delay in the acoustic transducer and in the electronic amplifiers and filters that we used.
As a validation test of the impulse response and phase methods for estimating velocity of sound in the lung, we determined velocity and dispersion of sound transmission in air. To do so, pseudorandom noise emitted from the sound stage was recorded by a microphone at measured distances from the stage. Using the impulse response method, we calculated the slope of the relationship between distance of the microphone from the sound stage against the delay in start of the impulse (r = 0.998); this yielded a velocity of 346 m/s, very close to the predicted velocity of 343 m/s at 20°C. Using the unpicked phase data, we estimated the phase velocity of sound at a number of frequencies: 8 kHz (346 m/s), 4 kHz (345 m/s), 2 kHz (360 m/s), and 1 kHz (368 m/s). Despite some divergence at the two lower frequencies, all values were within 7% of the expected value, consistent with sound propagation in a nondispersive medium.
Attenuation of sound in the lung.
To calculate attenuation attributable to the lung itself, we corrected for the spherical spreading of the sound wave with distance, using the inverse relationship between distance and sound pressure for a spherical wave source, and then determined the excess attenuation expressed in decibels per centimeter at each inflation with respect to the uninflated lung.
Statistics
All values are means ± SE. Differences between means were tested using a one-way ANOVA with post hoc, pairwise multiple comparison using a Tukey test, with P < 0.05 being taken as the critical level.
|
RESULTS
|
|---|
Studies were performed on the isolated lungs of eight fetal lambs of gestational ages ranging from 140 to 145 days (term = 147 days gestation). Fetal weight was 3.91 ± 0.18 kg, within the normal weight range for near-term fetuses. The calculated densities over which we made measurements ranged from 1.0 g/ml in the liquid-filled state to 0.35 ± 0.01 g/ml when the lung was maximally inflated with air; it appeared from visual observation that air injected into the lung distributed uniformly throughout the entire lung.
The electromagnetic sound transducer that we used had a relatively flat frequency response from 70 Hz to 7 kHz (Fig. 2, top trace). Progressively increasing the volume of air in the lung resulted in a shift to the left in the frequency spectrum of the noise transmitted through the lung (Fig. 2). This effect, particularly evident on the falling limb of the spectrum, is illustrated by the points superimposed on the individual traces in Fig. 2. By analogy with the O2 half-saturation pressure of blood, we defined the indicated value as f40, or the maximum frequency of sound transmitted through the lung at a sound pressure level of 40 dB. After pooling the data for all eight lungs, we found a rapid fall in f40 as lung density decreased, although most of the fall occurred from 1.0 to 0.5 g/ml (Fig. 3).
View larger version (19K):
[in this window]
[in a new window]
|
Fig. 2. Sound transmission, shown as sound pressure level (SPL), as a function of frequency in a lung as it was inflated with an increasing volume of air (from 0 to 240 ml). Note that the effect of inflation on the transmission spectra was particularly pronounced in the high-frequency sound range, as evidenced by the left shift in f40, the upper frequency of transmitted sound at a level of 40 dB. The spectrum of the input noise to the lung, derived from the noise signal recorded with an electret microphone placed directly over the orifice in the metal plate, is displayed as the topmost curve.
|
|
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 3. Relationship between f40 and density ( ) in the lung. Note that, at a density of 1 g/ml, high-frequency noise passed through the lung and that progressive aeration of the lung resulted in a rapid fall in the uppermost frequency crossing the lung. The best fit polynomial regression line is shown dashed. Note that f40 data were binned in 0.1 g/ml bins for each lung before the average and SE for each bin were calculated.
|
|
An example of the impulse response of the lung is shown in Fig. 4, top; note that the inset shows the method for specifying the start of the impulse response. The impulse response delay was corrected for a small residual time delay in our electronic equipment (measured using a microphone applied directly to the sound stage surface), and the result was then used to calculate the velocity of sound through the lung. For reasons discussed later, no reliable estimate of velocity was possible at a density of 1.0 g/ml. As shown by the upper set of data points in Fig. 4, bottom, as density decreased from 0.93 ± 0.01 to 0.75 ± 0.01 g/ml, there was a large fall in velocity of sound through the lung, from 187 ± 28.2 to 87 ± 3.7 m/s. Thereafter, as density fell to 0.35 ± 0.01 g/ml, there was a further decline in velocity to 69.6 ± 4.6 m/s. Included in Fig. 4, bottom, we have displayed two curves: the lower curve is the predicted velocity vs. density, using the relationship of Rice (16), and the upper curve is the predicted velocity times a factor of 3. As can be seen, the location of data points for velocity calculated from the impulse response appeared to conform with the shape of the Rice curve, but these velocities were approximately threefold greater than predicted by the Rice relationship. The lower set of data points in Fig. 4, bottom, shows the phase velocity determined for a frequency of 500 Hz. It can be seen that these conform reasonably closely to the velocity predicted from the Rice relationship.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4. Influence of decreasing density on the impulse response of the lung (top) and on the velocity of sound in the lung (bottom). Top: the impulse responses progressively shifted to the right as density decreased. Note the method used to determine the time of first arrival is shown in the inset. Bottom: two sets of data are plotted: the upper set represents velocity as determined from the impulse response delay, and the lower are estimates of the velocity of sound at a frequency of 500 Hz in the lung, derived from phase data. For both data sets, velocity estimates for each lung were binned in 0.1 g/ml density bins before the mean and SE were calculated. The lower dashed curve represents the predicted values of velocity vs. density based on the quadratic relationship published in Rice (16), whereas the velocity values predicted by that relationship when multiplied by a factor of 3 are shown by the upper dashed curve. Note that velocities estimated from the impulse response lie close to the upper curve, whereas the velocity at a frequency of 500 Hz, as calculated from phase, lies close to the theoretical curve.
|
|
Calculation of phase velocity for frequencies between 500 Hz and 5 kHz revealed a strong frequency dependence of sound velocity in the inflated lung (Fig. 5). This dispersive feature appeared to become more pronounced as density increased.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 5. Effect of lung inflation on the velocity of sound in the lung as a function of frequency, showing phase velocity calculated over the frequency range from 500 Hz to 5 kHz. Note the frequency-dependent or dispersive nature of sound transmission. In addition, it is clear that, as density falls, so also does the velocity of sound. Values are means ± SE.
|
|
The measurements of excess attenuation vs. inflation are illustrated in Fig. 6 for four frequencies over the octave range between 1,500 Hz and 3 kHz. Above 3 kHz, the SNR at high inflation was progressively reduced <20 dB, and, as a result, the data quality was insufficient to make accurate determination of attenuation, whereas <1,500 Hz there was no significant change in attenuation with inflation. The results at 1,500 Hz show a more or less linear increase in attenuation with inflation between 40 and 240 ml with (excess) attenuation rising significantly from 0.2 ± 0.5 to 3.6 ± 0.4 dB/cm (P < 0.05) over this inflation range. At 3 kHz, the rate of increase of attenuation with inflation is initially higher than at 1,500 Hz, rising from –0.2 ± 0.6 at 40 ml to 9.5 ± 1.7 dB/cm at 120 ml, but, thereafter, there was no dependence on inflation. Comparing values at an inflation of 80 ml, the attenuation at 3 kHz (6.0 ± 1.8 dB/cm) was significantly higher (P < 0.05) than at 1,500 Hz (1.2 ± 0.4 dB/cm), indicating a strong frequency dependence of attenuation at constant inflation.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 6. High-frequency attenuation in dB/cm during inflation of the fetal lung showing the effect of increasing frequency. Values are means ± SE.
|
|
|
DISCUSSION
|
|---|
Two principal findings emerge from this study. First, progressive inflation of the lung of the near-term sheep fetus with air results in a pronounced decline in the transmission of audible sound through the lung. Attenuation of sound, which occurs across the entire audible range, is particularly pronounced at high frequencies as lung density decreases from 1.0 to 0.5 g/ml, after which a more gradual attenuation occurs until density reaches 0.35 g/ml, the minimum we could achieve without rupturing the lung. Second, we show that the velocity of sound transit through the lung declines precipitously when lung density decreases from 1.0 to 0.7 g/ml and, thereafter, falls more gradually down to a density of 0.35 g/ml. The acoustic properties of the newborn lung suggest sound in the audible range has potential as a means for monitoring the degree of lung inflation; in turn, knowledge of lung inflation could be a valuable aid in tailoring ventilation strategies to suit the individual infant’s needs.
Our finding of a decrease in high-frequency sound transmission through the lung as it is progressively aerated is similar to reports in the isolated pig lung (8) and in the human (14). It is also in accord with the experimental evidence that the attenuation of sound introduced to the lung decreases when the lung is made edematous (4, 13, 15) and the well-known clinical observation that an increase in transmission of high-frequency sound occurs, giving rise to auscultatory signs such as "whispering pectriloquy" when the lung is consolidated in the presence of pneumonia (19). Furthermore, the strong frequency dependence of normalized attenuation indicates that the increase in attenuation with inflation at high frequencies is an intrinsic property of the lung tissue and not simply a manifestation of increased distance traveled by the sound waves. The high velocity of sound that we measured in the liquid-filled lung is consistent with the known velocity of sound in soft tissues (1,500 m/s), and the rapid fall in velocity once air enters the lung is predicted by theory (16). Furthermore, previous measurements have shown that sound velocity falls as lung inflation increases across the normal physiological range (8, 11, 16).
Critique of Methods
The fetal lung was selected for these studies because it offered the opportunity to examine the acoustic properties of the lung, particularly over the very low volume range. In our experience, it is difficult to achieve uniform aeration in a degassed lung, whereas we knew from past visual observation that the liquid-filled fetal lung appears to expand uniformly when air is gradually introduced into it. With this preparation, we have gained insight into the acoustic properties of the lung at very low levels of inflation and high lung densities. These observations are of particular relevance to the preterm infant with respiratory distress syndrome in whom surfactant deficiency commonly results in very low lung volumes and diffuse atelectasis. A drawback of the model is that the presence of fetal lung liquid did not permit the achievement of very low lung densities, so we were unable to examine sound velocity and attenuation over the range of densities expected when the lung is significantly overdistended. There is, however, good theoretical and experimental evidence to indicate that the velocity of sound in the lung increases considerably as its density falls toward 0 g/ml (16), suggesting that the acoustic technique may be a useful means for detecting overdistension of the lung as well as underinflation.
A further limiting feature of our study is the acoustic data-acquisition software, which had a maximum sampling frequency of 48 kHz and a corresponding minimum sample interval of 0.02 ms. With sound crossing soft tissue at 1,500 m/s, sound is expected to cross an 2-cm depth of the fetal lung within 0.013 ms, close to the sampling period. Thus our estimated velocities incorporate a quantized error that is large when velocity is high in the liquid-filled lung. In view of the size of the quantized error, we did not make an estimate of velocity in the liquid-filled lung; the size of the error is reflected in the large standard error of the velocity estimate at a density of 0.93. However, once lung density is <0.9 g/ml, and velocity has fallen to 100 m/s (Fig. 4), sound would cross a 2-cm-thick lung in 0.2 ms, reducing the quantized error of estimates to <10%. When this velocity was reached in our study, the lung was >2 cm thick because of its greater inflation, and the increased path length would further reduce the error of the estimated velocity.
A key aspect of our study design was to introduce sound directly to the lung surface, rather than via the airway, the route of entry for sound in most earlier studies using isolated lungs (8, 17) or using intact animals (4, 15) and humans (1, 6, 10). Although studies utilizing airway entry of sound report that lung inflation has an effect on the velocity of sound from its point of entry at the airway opening to the surface of the lung or chest wall, interpretation of results requires assumptions about the pathway that sound takes through the respiratory system (1, 8). The possibility that low and high frequencies do not take the same path (12, 20) further complicates the interpretation of results. As others have recognized (11), by introducing sound directly to the lung surface, we can be confident of achieving purely parenchymal passage of sound, thereby simplifying the interpretation of our data.
A recent study in which sound was introduced via the supraclavicular space to the human lung showed that velocity of sound in the lung increases across the volume range from residual volume to total lung capacity (11), confirming earlier findings in the isolated horse lung (16). Our results in the fetal sheep lung show the opposite, a decline in velocity with inflation. These apparently conflicting findings are largely reconciled by the theoretical quadratic relationship between velocity and density (16). The relationship predicts a U-shaped velocity vs. density curve (see Fig. 4, bottom), with high velocity in the solid or liquid-filled state (1,500 m/s), declining rapidly to a minimum velocity at 0.5 g/ml at the center of a relatively flat portion of the curve before rising abruptly in the high-inflation, low-density condition. Note that, as density approaches that of air, velocity should theoretically approach 73 m/s. However, sound velocity in the isolated fetal lung continued to fall below a density of 0.5 g/ml. We ascribe this effect to the dispersive nature of sound transmission in the lung, with further inflation at densities <0.5 g/ml, resulting in further loss of the higher frequency and faster traveling sound waves in our input noise signal. In support of this contention, when the influence of dispersion was removed by plotting phase velocity at a constant frequency (see Fig. 4, bottom), this effect is not apparent.
Of interest in our study is the approximately threefold higher velocity that we calculated from the impulse response in the isolated fetal sheep lung compared with that in the isolated horse lung (16) and in the human lung (11). On the basis of the theory outlined by Rice (16), which assumes that the lung is a two-phase system, with the alveoli not communicating directly with one another (closed-cell model), it is lung density (or gas fraction) and gas stiffness that determine velocity. This theory predicts that the minimum velocity (24 m/s) occurs at a density of 0.5 g/ml, a value achieved in the horse (16) and the current study and likely to have been achieved in the human (11). One possible explanation for differences in estimated velocity is uncertainty about the precise time of arrival of introduced sound at the recording device. Figure 4, top (inset), shows that the beginning of the impulse response in our study can be specified with some precision, so that our estimate of velocity is reasonable. Rice (16) did not include an example of an impulse response in his report. Paciej et al. (11) used a short burst of narrow-bandwidth frequency modulated tone for velocity estimates, which appears to provide a reliable means of specifying the delay between input and transmitted sound. Another possibility is that the closed-cell model, on which theoretical predictions of velocity are based, does not apply across species and across inflation states within a species. Indeed, Rice (16) noted a large discrepancy in measured and theoretical velocities in the horse lung at high volumes, which he suggested could be ascribed to interalveolar communication. That is, the cells, or alveoli, are not closed at high volumes: perhaps the 140-day-old fetal sheep has a relatively open cell structure, resulting in the higher velocities that we measured compared with theory.
A more likely explanation for the discrepancy between our velocity estimates and published values is suggested by our analysis of phase data. With this approach, we observe a strong frequency dependence or dispersion of sound velocity in the lung, with higher frequencies traveling faster than low frequencies. We suggest that the impulse response method of estimating velocity of sound in the lung using the first arrival time provides a value that is dominated by the highest frequencies transmitting and, therefore, biases the result toward high velocities. If band-limited measurements of the impulse response are made, either deliberately or inadvertently, lower values for the velocity would be expected if the lung is a dispersive medium. This limitation might be caused by frequency-dependent microphones or acoustic transducers or even the state of inflation of the lung. Consistent with this explanation, Paciej and coworkers (11) measured very low velocities at 150 Hz, whereas our results, which used frequency content extending up to 7 kHz, show much higher velocities. Rice (16) noted that, after the impulse he delivered to the horse lung had propagated through the lung, there was little energy in the recorded signal at frequencies >1 kHz, suggesting that band limitation of the broad spectrum impulse noise by the lung, or by his recording equipment, may have led to the lower velocity values that he recorded. In support of this suggestion, there is close agreement between velocity of sound in the lung at 500 Hz calculated from our data and the theoretical values derived from Rice (16).
In recent years, there has been growing recognition that maintenance of an optimal lung volume during mechanical ventilation plays an important role in minimizing ventilator-induced lung injury (2, 18). Strategies to avoid injury from overinflation of the lung and underinflation of the lung are limited by lack of any suitable techniques by which the degree of lung inflation can be continuously monitored in patients on mechanical ventilator support. Our study suggests that two readily measured acoustic variables, the attenuation of audible sound within the lung and the velocity of sound as it crosses the lung, are sensitive to the degree to which the isolated lung of the fetal sheep is inflated with air. Thus our findings and the earlier work of Rice (16) demonstrate that these two variables offer potential for monitoring the degree of lung aeration, at least in isolated lungs, with particular sensitivity at the extremes of lung volume.
|
GRANTS
|
|---|
This research was supported by the National Health and Medical Research Council of Australia and by PulmoSonix Pty. Ltd., which funded the research through a contract with Monash University.
|
DISCLOSURES
|
|---|
P. J. Berger, C. A. Ramsden, and M. H. Wilkinson are named as Inventors on patents relating to the use of acoustics in the respiratory system, and they are shareholders in PulmoSonix Pty Ltd.
|
ACKNOWLEDGMENTS
|
|---|
The authors acknowledge the skilled technical assistance provided by Peter Camilleri.
|
FOOTNOTES
|
|---|
Address for reprint requests and other correspondence: P. J. Berger, Ritchie Centre for Baby Health Research, Monash Institute of Reproduction and Development, Monash Medical Centre, Level 5, 246 Clayton Rd., Clayton, 3168, Australia (E-mail: philip.berger{at}med.monash.edu.au)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
REFERENCES
|
|---|
- Bergstresser T, Ofengeim D, Vyshedskiy A, Shane J, and Murphy R. Sound transmission in the lung as a function of lung volume. J Appl Physiol 93: 667–674, 2002.
- Clark RH, Gerstmann DR, Jobe AH, Moffitt ST, Slutsky AS, and Yoder BA. Lung injury in neonates: causes, strategies for prevention, and long-term consequences. J Pediatr 139: 478–486, 2001.
- Clark RH, Slutsky AS, and Gerstmann DR. Lung protective strategies of ventilation in the neonate: what are they? Pediatrics 105: 112–114, 2000.
- Donnerberg RL, Druzgalski CK, Hamlin RL, Davis GL, Campbell RM, and Rice DA. Sound transfer function of the congested canine lung. Br J Dis Chest 74: 23–31, 1980.
- Dreyfuss D, Basset G, Soler P, and Saumon G. Intermittent positive-pressure hyperventilation with high inflation pressures produces pulmonary microvascular injury in rats. Am Rev Respir Dis 132: 880–884, 1985.
- Kiyokawa H and Pasterkamp H. Volume-dependent variations of regional lung sound, amplitude, and phase. J Appl Physiol 93: 1030–1038, 2002.
- Kraman SS. Speed of low-frequency sound through lungs of normal men. J Appl Physiol 55: 1862–1867, 1983.
- Leung A, Sehati S, Young JD, and McLeod C. Sound transmission between 50 and 600 Hz in excised pig lungs filled with air and helium. J Appl Physiol 89: 2472–2482, 2000.
- Lu S, Doerschuk PC, and Wodicka GR. Parametric phase-delay estimation of sound transmitted through intact human lung. Med Biol Eng Comput 33: 293–298, 1995.
- Mahagnah M and Gavriely N. Gas density does not affect acoustic transmission in normal men. J Appl Physiol 78: 928–937, 1995.
- Paciej R, Vyshedskiy A, Shane J, and Murphy R. Transpulmonary speed of sound input into the supraclavicular space. J Appl Physiol 94: 604–611, 2003.
- Patel S, Lu S, Doerschuk PC, and Wodicka GR. Sonic phase delay from trachea to chest wall: spatial and inhaled gas dependency. Med Biol Eng Comput 33: 571–574, 1995.
- Ploysongsang Y, Michel RP, Rossi A, Zocchi L, Milic-Emili J, and Staub NC. Early detection of pulmonary congestion and edema in dogs by using lung sounds. J Appl Physiol 66: 2061–2070, 1989.
- Pohlmann A, Sehati S, and Young D. Effect of changes in lung volume on acoustic transmission through the human respiratory system. Physiol Meas 22: 233–243, 2001.
- Räsänen J and Gavriely N. Detection of porcine oleic acid-induced acute lung injury using pulmonary acoustics. J Appl Physiol 93: 51–57, 2002.
- Rice DA. Sound speed in pulmonary parenchyma. J Appl Physiol 54: 304–308, 1983.
- Rice DA and Rice JC. Central to peripheral sound propagation in excised lung. J Acoust Soc Am 82: 1139–1144, 1987.
- Slutsky AS. Lung injury caused by mechanical ventilation. Chest 116: 9S–15S, 1999.
- Welsby PD and Earis JE. Some high pitched thoughts on chest examination. Postgrad Med J 77: 617–620, 2001.
- Wodicka GR, Aguirre A, DeFrain PD, and Shannon DC. Phase delay of pulmonary acoustic transmission from trachea to chest wall. IEEE Trans Biomed Eng 39: 1053–1059, 1992.
- Wodicka GR and Shannon DC. Transfer function of sound transmission in subglottal human respiratory system at low frequencies. J Appl Physiol 69: 2126–2130, 1990.
Copyright © 2005 by the American Physiological Society.
TOP
ABSTRACT
MATERIALS AND METHODS
