INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS AND PROTOCOL RESULTS DISCUSSION APPENDIX REFERENCES

NONINVASIVE EVALUATION of airway geometry and the mechanical properties of the human respiratory system has remained a constant challenge for clinicians and engineers, especially in spontaneously breathing subjects. Two distinct methods, based on analysis of small, time-varying external forces, are currently available to perform this type of geometric/mechanical assessment: 1) the acoustic reflection method, initially proposed in humans by Fredberg et al. (10), which provides the airway geometric properties, and 2) the forced-oscillation method, first described by Dubois et al. (6), which provides the mechanical properties of the entire respiratory system. In present applications, these two methods use two distinct types of excitation signals, i.e., a transient pulse for the acoustic reflection method and a steady state of periodic oscillations or pseudorandom noise input for the forced-oscillation method, making concomitant assessment of geometric/mechanical properties difficult. An obvious improvement would therefore consist of simultaneous application of the two external forcing methods to combine the previously demonstrated clinical benefits of each into a single method. Briefly, the acoustic reflection method provides a spatial representation of proximal airways, whereas the forced-oscillation method provides the mechanical properties of the entire respiratory system, including distal airways, but without providing precise information about their spatial distribution. Furthermore, the acoustic reflection method not only allows oral, nasal, and tracheal airway assessments (16), but recent miniaturization of the recording apparatus has also allowed this method to be applied in uncooperative, intubated intensive care patients (18, 24). The forced-oscillation method has been predominantly used in nonintubated patients rather than intubated subjects.

To simultaneously apply the acoustic reflection method and the forced-oscillation method, we propose dual analysis of a single transient wave to allow concomitant assessment of the geometric and mechanical properties of the respiratory system. This new approach, together with the associated theory derived from wave propagation equations, described below, constitutes a new method dedicated to dual assessment of upper airway area and respiratory input impedance (called AAI).

	METHODS AND PROTOCOL

TOP ABSTRACT INTRODUCTION METHODS AND PROTOCOL RESULTS DISCUSSION APPENDIX REFERENCES

Methods

Area profile. The acoustic reflection method, which gives the longitudinal cross-sectional area profile along the airway, has already been described, and its limits have been thoroughly defined for the various living applications (3, 7, 10, 14, 17, 24). This method is based on analysis of a planar acoustic wave propagating in a rigid duct connected to the airway. Briefly, oscillatory pressure and oscillatory flow rate in a duct can both be described by the sum of two waves propagating in opposite directions with the same wave speed (20)

(1)

(2)

where P is the pressure, is the flow,  is the pulsation, and x is the spatial coordinate. P_i, the incident wave, is the pressure wave propagating in the positive x direction, whereas P_r, the reflected wave, is the pressure wave propagating in the negative x direction. Zc, called characteristic impedance, reflects the relationship between pressure and flow waves and tends toward ( · c)/A(x) with increasing values of pulsation where c is the wave speed, A is the cross-sectional area, and  is the gas density. Using this equation system, written in the time domain, Ware and Aki (26) showed that A(x) can be inferred from the value of the impulse response under a set of restrictive conditions. This impulse response, H, is defined in the frequency domain as

(3)

Previous studies have shown that measurement of the impulse response can be used to infer the area-vs.-distance function from the mouth to the end of the trachea (10), from the nostril to the sinuses (15), or from the ventilator connector to the distal end of an endotracheal tube (ETT) (24).

Impedance. The impulse response can be used to obtain the input impedance at the entry of the airways according to the transitory forced-oscillation method described by Fredberg et al. (10). The impedance, Z, is classically defined in the frequency domain as the ratio between pressure and flow. Z can be computed from Eqs. 1-3

(4)

If a wave tube is connected to the respiratory system, the value of H at the end of the wave tube can be used to compute the respiratory impedance by Eq. 4. Equation 4 does not require any assumption concerning the system measured, which simply corresponds to the load impedance at the end of the wave tube. Respiratory impedance, computed from the impulse response and wave-tube characteristics, therefore, corresponds to the overall result of the entire respiratory system, including distal airways, tissues, and chest wall properties.

(5)

(6)

Setup. The experimental device used to measure area and impedance consisted of a wave tube with two microphones (piezoresistive pressure transducers 8510-B; Endevco France, Le Pré-Saint-Gervais, France) and a horn driver (Fig. 1) tightly connected to the proximal airway opening. A transient acoustic wave was generated by the horn driver driven by a personal computer via a digital-to-analog converter. The resulting pressure in the wave tube was recorded via the microphones for a period of ~0.2 s. The peak-to-peak amplitude of the oscillating pressure was ~1 cmH₂O. The frequency content of the pressure included the range 10-4,500 Hz. Microphone outputs were fed to an analog-to-digital converter (14 bits, 24-µs sampling period, E. Benson Hood Laboratories, Pembroke, MA) and recorded on a personal computer. The discretized pressures (8,192 points/channel) were converted into the frequency domain and analyzed as indicated above to infer impulse response, area, and input impedance. The choice of recording procedure, i.e., sampling frequency and number of points recorded, introduced a spatial increment of ~0.4 cm for A(x) and a frequency increment of ~5 Hz for Z(). On each day of measurement, the two transducers were self-calibrated, as indicated in the APPENDIX.

View larger version (14K): [in this window] [in a new window]   Fig. 1.   Diagram of the wave tube. Pressure, the sum of the incident and reflected waves, is recorded at two loci of the wave tube (x = L and x = 0, where x is the spatial coordinate and L is distance) to infer impulse response, longitudinal airway area profile, and input impedance vs. frequency. ETT, endotracheal tube.

Protocol

The ability of AAI to estimate impedance beyond a given distance of ETT, as indicated by Eq. 6, was tested by measuring the impedance of a known model (model 4) with the 0.8-cm-diameter wave tube. This model consisted of an 8-mm ETT connected to a circular tube with a constant diameter. Impedance was measured either from the site of the second microphone, as indicated in Eq. 6, or immediately beyond the distal extremity of the ETT, by using Eqs. 5 and 6.

Moreover, because the frequency response of these models was very different from that of the physiological model, at least in the low- to medium-frequency range (e.g., the real part is much lower than the imaginary part, in contrast with airway impedance), we constructed a physical model with a frequency response resembling the physiological response. We used a Plexiglas tube (length 10 cm and inner diameter 1.55 cm) filled with a parallel network of small capillaries made of undulated aluminum foil connected to the inner volume of a rigid spherical chamber (inner volume 1,725 cm³, inner diameter ~15 cm). This resulted in an increase in the real part of the impedance at the lower frequency range (<100 Hz) associated with a negative imaginary part at the lowest frequency. The distal end of a 7-mm ETT was introduced into the Plexiglas tube. The ETT cuff was used to obtain a good seal between the ETT and Plexiglas tube. ETT obstructions due to mucus deposition were also simulated in the 7-mm ETT by inserting a full cylinder into the ETT. Moderate obstruction was simulated by using a cylinder 6.8 cm long and 0.07 cm² in cross-sectional area placed ~4.5 cm from the distal end of the ETT, whereas major obstruction was induced by another cylinder, 3 cm long and 0.20 cm² in cross-sectional area, placed ~5 cm from the distal end of the ETT. Because most forced-oscillation studies allowing human diagnostic applications, e.g., evaluation of mechanical nonhomogeneity and effect of drugs such as histamine (2, 19, 21, 23, 27), were performed by using more or less implicitly lumped-parameter models, we limited the frequency range of interest up to 90 Hz for the physiological model, at which point the validity of lumped parameters starts to be questionable. Above this frequency, the length of the respiratory system may no longer be negligible in relation to the pressure wavelength, and this model may no longer be more physiologically relevant than other rigid-wall long-tube models used in this study (models 1, 2, 3, and 4 in Table 1). The results obtained with the physiologically relevant physical model were compared with a "reference" method that consisted of directly measuring the model without the ETT in between the wave tube and the model.

The AAI method was also tested in two intubated patients in the Henri Mondor Hospital intensive care unit. Measurements were performed in the context of a protocol approved by the Ethics Committee of the Société de Réanimation de Langue Française (French Society of Intensive Care Medicine). Informed consent was obtained from the patients or their closest relatives. One patient had a history of chronic obstructive pulmonary disease (COPD), whereas the other was intubated with no respiratory disease (no-RD). The patients were briefly disconnected from the ventilator and breathed spontaneously during the measurement, as routinely performed during an endotracheal suctioning procedure. One measurement consisted of 10 pulses (total duration ~2 s). Each pulse was analyzed separately in terms of area vs. distance and impedance vs. frequency.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS AND PROTOCOL RESULTS DISCUSSION APPENDIX REFERENCES

Figure 2 presents the results obtained with the AAI method in two mechanical models (models 1 and 2 in Table 1) using the 1.9-cm-diameter wave tube. The inner diameters of these tubes (wave tube and models) roughly correspond to oral applications in adults. The areas inferred with the AAI method were in agreement with real values (Fig. 2A), i.e., a first 2.8-cm² tube followed by a second 2-cm² tube for model 1 and a 2.8-cm² tube closed at its end for model 2. The impedance values measured over the entire frequency range (10-2,000 Hz) were close to the reference values for both the real and imaginary parts (Fig. 2B) for model 1, whereas a marked discrepancy was observed between reference and measured values with model 2 at the lowest frequency.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Area profile and input impedance of models 1 and 2 measured with the 1.9-cm-diameter wave tube. A: area vs. distance for these 2 models. Model 1 (solid line) is a 5.4-cm-long tube with a 2.8-cm2 constant area connected to a 114-cm-long tube with a 2.0-cm2 constant area open to the atmosphere. Model 2 (dashed line) is a 24.5-cm-long tube with a 2.8-cm2 constant area closed with a rigid stopper. B: real and imaginary parts of impedance (Z) of model 1 measured at x = 0, Z(0) (), compared with the analytical solution (1, 4) of the oscillatory compressible flow (solid line) considered to be the reference method. C: same as B but for model 2.

The results obtained with the AAI method (in model 3) with the 0.8-cm-diameter wave tube, i.e., the wave tube dedicated to the ETT application, were very similar to those observed with the 1.9-cm wave tube. Areas and impedance measured with the AAI method were still in agreement with theoretical values (Fig. 3).

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Area profile and input impedance of model 3 measured with the 0.81-cm-diameter wave tube. A: area vs. distance of model 3. The model is an 8-cm-long tube with a 0.5-cm2 constant area connected to a 59.8-cm-long tube with a 0.7-cm2 constant area open to the atmosphere. B: real and imaginary parts of impedance of model 3 measured at x = 0, Z(0) (), compared with the analytical solution (1, 4) of the oscillatory compressible flow (solid line) considered to be the reference method.

To describe the effect of the locus used to estimate the impedance for ETT applications, we compared the impedance measured at the end of an 8-mm ETT in model 4 (Fig. 4). Only impedances measured from the distal end of the ETT, i.e., by Eqs. 5 and 6, were in agreement with the theoretical values of these models over the entire frequency range tested. In particular, the frequencies for which a relative maximum in terms of impedance was predicted by the reference method were correctly observed when the impedance was measured at the distal end of the ETT. In contrast, the impedance measured at the second microphone, located close to the proximal end of the ETT, Z(0), was very different from the theoretical value.

View larger version (28K): [in this window] [in a new window]   Fig. 4.   Evaluation of the dual assessment of upper airway area and respiratory input impedance (AAI) method to estimate impedance beyond a given distance of ETT. A: area vs. distance of an 8-mm ETT connected to a 17.4-cm-long tube with a 2.8-cm2 constant area open to the atmosphere. B: real and imaginary parts of impedance, estimated by the AAI method at the proximal end of the ETT, Z(0) (+), and at the distal end of the ETT, Z(ETT end) (), were compared with the analytical solution of the 17.4-cm-long tube (1, 4) (solid line) considered to be the reference method.

Evaluation of the AAI method performed in the physiologically relevant physical model (see Protocol, above) is presented in Fig. 5 for a 7-mm ETT. Longitudinal area profiles were closely correlated with the expected area profiles (Fig. 5A). Without obstruction inside the 7-mm ETT, the area of the ETT was found to be ~0.4 cm². When full cylinders with cross-sectional areas of 0.07 and 0.2 cm² were successively introduced inside the 7-mm ETT to simulate ETT obstruction, the resulting partial obstructions were detected, quantified, and located (Fig. 5A). The 0.07-cm² cylinder induced a moderate reduction in ETT area (20%), whereas the 0.2-cm² cylinder induced a major reduction (50%). For all obstruction conditions, i.e., no, moderate, or major obstruction, the impedance measured at the second microphone [Z(0); Fig. 5B] clearly differed from the impedance of the physical model obtained by the "reference" method, i.e., the AAI method without an ETT between the wave tube and the model (see Protocol). In contrast, for no obstruction and moderate obstruction, the impedance measured at the distal end of the ETT, Z(ETT end), was fairly close to that measured with the reference method (Fig. 5B). Under conditions of major obstruction, the real part of impedance measured at the distal end of the ETT was slightly overestimated compared with the reference method, whereas the imaginary part was in agreement with the reference method.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Effect of obstruction on estimation of impedance at the distal end of the ETT by the AAI method. A: area vs. distance of a 19.5-cm-long 7-mm ETT without obstruction, with moderate obstruction and with major obstruction. Moderate obstruction was simulated with a cylinder 6.8 cm long and 0.07 cm2 in cross-sectional area placed ~4.5 cm from the distal end of the ETT, whereas major obstruction was simulated with another cylinder 3 cm long and 0.20 cm2 in cross-sectional area placed ~5 cm from the distal end of the ETT. B: real and imaginary parts of the more physiological model estimated at the proximal end of the ETT, Z(0) (solid symbols), and at the distal end of the ETT, Z(ETT end) (open symbols), under all conditions of obstruction, i.e., no obstruction (diamonds), minor obstruction (triangles), and major obstruction (circles). These impedance measurements are compared with the results of the AAI method without ETT between the wave tube and the model considered to be the reference method (solid line).

Evaluation of the AAI method in two intensive care patients (COPD and no-RD; see Protocol) showed that the ETT of the COPD patient was not obstructed, whereas the ETT of the no-RD patient was slightly obstructed near its distal end (Fig. 6A). In these patients, estimation of impedance at the end of the ETT, Z(ETT end), compared with estimation at the second microphone, Z(0), reduced the frequency dependency of the real part of impedance, at least in the clinically relevant frequency range (Fig. 6B).

View larger version (26K): [in this window] [in a new window]   Fig. 6.   Examples of area profile and impedance in two intubated patients measured in the Henri Mondor Hospital intensive care unit. Both patients were intubated with an 8-mm ETT, and measurements were performed with an 8-mm-diameter wave-tube diameter. One patient had a history of chronic obstructive pulmonary disease (COPD), whereas the other was intubated with no respiratory disease (no-RD). A: area profile of the ETTs. B: real and imaginary parts of impedance estimated at the proximal end of the ETT, Z(0) (solid symbols) and at the distal end of the ETT, Z(ETT end) (open symbols). Error bars indicate the standard deviation associated with a series of ten pulses.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS AND PROTOCOL RESULTS DISCUSSION APPENDIX REFERENCES

It is generally accepted that oscillation methods constitute powerful tools to explore the respiratory system. Together with modeling and parameter estimations, these methods have been shown to be valuable to detect and interpret respiratory system abnormalities associated with disease. However, clinical applications, especially in the intensive care unit, have remained limited for various reasons, including the considerable measuring time compared with the respiratory time required for the forced-oscillation method. The acoustic reflection method setup has also been too cumbersome for intensive care unit applications, before the miniaturization proposed by Louis et al. (18). The present study demonstrates that a single transient wave presents a sufficient frequency content to assess both airway geometry and respiratory input impedance from analysis of its reflection. Our approach can be considered to be a further improvement of both oscillation methods, in which the two methods are combined in a single method, AAI, with 1) simultaneous measurement of area profile and respiratory impedance and 2) a miniaturized setup that has already been shown to be valuable for intensive care applications.

Physical implications. Measurements of input impedance inferred with the AAI technique were in agreement with the theoretical predictions from 10 to 2,000 Hz except at low frequencies (10-50 Hz) for model 2. One possible reason for these discrepancies could be an end effect due to the stopper because the assumption of perfect lossless reflection at the stopper, used to compute the impedance of the model, is clearly an oversimplification of the problem. Our stopper was made of Plexiglas, which does not have an infinite acoustic impedance. Other possible reasons concern the same limitation observed with the wave-tube method (8) to measure very high impedance values at low frequencies (9). Briefly, Eq. 4 can be recast as:

Input impedance at the end of the ETT. The ETT area profile inferred by acoustic reflection can be used to estimate, by the incremental procedure described in Eq. 6, the input impedance from the distal end of the ETT, provided that the ETT retains a circular shape (Figs. 4 and 5). This property remains true when the shape of the ETT is moderately altered, i.e., when the area variation is <20% (Fig. 5). In the case of a marked alteration of shape, such as the major ETT obstruction simulated in this study (Fig. 5), the diameter derived from the area measurement can no longer be used to compute a Witzig-Womersley number, , representative of the viscous and thermal loss. Basically,  represents the ratio between tube radius and the viscous boundary layer (1, 8). Altering the circular shape of a tube tends to modify the ratio between tube radius and viscous boundary layer, i.e., . For high values of  ( > 5), the viscous loss, i.e., the tube resistance, increases with . In contrast, inertia and compliance of the gas tend to become constant with increasing values of . Consequently, altering the circular shape of a tube tends mainly to modify tube resistance, without severely altering either inertia or compressibility of the gas. This may explain why impedance measured at the end of an ETT with a major obstruction overestimated the real part of impedance, at least at low frequencies, whereas the imaginary part was correctly estimated (Fig. 5).

Physiological implications. Estimating respiratory impedance from the distal end of the ETT by the AAI method is clinically important because it provides a representative index of the patient's respiratory system, which is not affected by the geometric characteristics (diameter and length) of the ETT that vary from one patient to another. The AAI method provides respiratory system impedance from the end of the ETT only if the ETT is not obstructed or only moderately obstructed (20%). However, for measurements of the ETT area profile, the AAI method easily detects a major obstruction, allowing the concomitant altered respiratory system impedance measurement to be disregarded. It should be noted that the major obstruction tested in Fig. 5 is a clinically unacceptable situation because, during a trial of weaning, this degree of ETT obstruction would dramatically increase the patient's work of breathing with a risk of failure of the weaning procedure (5, 24, 25). Similarly, during controlled mechanical ventilation, a major ETT obstruction remains clinically unacceptable because it increases the risk of total ETT obstruction with possibly fatal consequences. In both of these clinical situations, a therapeutic procedure (suction maneuver, ETT extubation and reintubation with a new ETT, etc.) would be performed to restore ETT patency and normal ventilatory conditions.

1

1