A resonance-mode piezoelectric device for measurement of respiratory mechanics

doi:10.4236/jbise.2013.611133

Paper Menu >>

Journal Menu >>

J. Biomedical Science and Engineering, 2013, 6, 1062-1071 JBiSE

http://dx.doi.org/10.4236/jbise.2013.611133 Published Online November 2013 (http://www.scirp.org/journal/jbise/)

A resonance-mode piezoelectric device for measurement of

respiratory mechanics

Hamed Hanafi Alamdari, Lucas Posada, Swati A. Bhatawadekar, Jeremy A. Brown,

Geoffrey N. Maksym

School of Biomedical Engineering, Dalhousie University, Halifax, Canada

Email: [email protected], [email protected]

Received 10 September 2013; revised 15 October 2013; accepted 26 October 2013

License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

ABSTRACT

This paper presents the design o f a nove l oscillometry

device for the measurement of respiratory mechanics

based on piezoelectric bimorph actuator technology.

To predict performance for measurement of human

respiratory mechanics, a dynamic model was devel-

oped based on a bimorph piezoelectric actuator driv-

ing a linear resistance mesh screen including subject’s

respiratory impedance loads, and realistic breathing

noise. Model performance was also validated in a pro-

totype device. We found that while breathing noise

substantially lowered SNR, the model could produce

sufficient pressure and flow for acceptable SNRs ex-

ceeding 35 dB, and accuracies exceeding 99%. Sat-

isfactory accuracy could be achieved with load impe-

dance errors less than 3%. Maintaining the air-gap

around the oscillating mesh with a resistance against

the leak greater than 0.38 cmH2O/L/s maintained good

performance, with an acceptable 4 dB decrease to

SNR. Moreover, this work provides multiple solutions

to host higher amounts of noise and nonlinearities.

These results indicate that the development of an ac-

curate lightweight portable single frequency FOT de-

vice is feasible.

Keywords: Forced Oscillation Technique FOT;

Respiratory Mechanics; Tidal Volume

1. INTRODUCTION

Oscillometry also known as the Forced Oscillation Tech-

nique (FOT) superimposes fluctuations in airway pres-

sure on normal breathing to measure the mechanical im-

pedance to airflow of the respiratory system for diagnos-

ing and monitoring lung diseases [1]. Using FOT is pos-

sible in very young children to elderly patients as it is

easy to perform unlike the current standard approach,

which involves a learned maximal forced expiration that

is difficult for many patients. Respiratory impedance is a

complex quantity with a real part: respiratory resistance

(Rrs) largely due to airflow resistance and an imaginary

part: reactance (Xrs) arising from elastic properties of the

lung and chest wall and the inertia of the air. The resis-

tance is elevated in diseases such as asthma and COPD

associated with airway narrowing and the reactance can

also be altered due to loss of airspaces as airways close

and limit flow [2].

FOT Devices must reliably produce oscillatory wave-

forms even when perturbed by breathing, and must ac-

curately measure pressure and flow due to the oscilla-

tions, over a wide range of patient respiratory imped-

ances, and compensate for any self-impedance of the de-

vice and disposable filter. Current FOT devices use ei-

ther a loudspeaker or oscillating electromagnetic actuator

which results in a device that is larger or heavier than ty-

pical spirometers.

Piezoelectric actuators have been used for several years

in applications such as loudspeakers, mechanical damp-

ers, ultrasonic motors, precision position controlling, noise

control, relays, phonograph pick-up, acoustics, and pres-

sure sensing [3-6]. We have designed a smaller and less

costly FOT device, based on a piezoelectric bimorph ac-

tuator using a computer model to help explore the feasi-

bility and design requirements for a suitable device. The

design was tested in a prototype device.

2. METHODS

2.1. Simulations

The device is based on a moving mesh in a flow tube and

was modeled in MATLAB/SIMULINK as described in

following paragraphs. The resonance of the oscillating

cantilever was tuned to 6 Hz by varying the mass on the

tip [7]. Forced air oscillation at 6 Hz was sent into a res-

OPEN ACCESS

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071 1063

piratory system that modeled different subject popula-

tions including healthy, COPD, normal children and asth-

matic children with representative impedance values indi-

cated in Table 1 [8-11]. We used actual measured breath-

ing waveforms of 6 voluntary subjects (without any ap-

paratus) as realistic noise. Subjects were recruited as part

of a research ethics approved study approved by the Ca-

pital Health Authority (Halifax, NS). The average RMS

amplitude of the breathing flow was used as the noise

gain of unity. The noise was adjusted in amplitude as a

perturbing input to assess the effects of breathing noise

on predicted values as well as on the actuator. The tidal

volume (VT) of the measured volunteer patients was 660

± 150 mL (±SD) while 1/3 of that was used as realistic VT

for the child patient [12]. The predicted actuation char-

acteristics for piezoelectric bimorphs are well known [5]

and therefore to achieve sufficient accuracy, signal-to-

noise ratio (SNR), low power consumption, and low cost,

the model incorporated a novel resonance based piezo-

electric actuator design. To model the leak resistance of

the moving mesh screen, the mesh resistance comprising

the holes on the mesh and the air-gap with the housing

was modeled as a resistance in parallel to the respiratory

system impedance. The device needs to reach high air-

gap resistance to conduct the least amount of air loss.

From the model we set a design goal for a minimum air-

gap threshold resistance to achieve errors less than 10%

[13] for different subject loads. Differences between

groups due to the influence of breathing noise on SNR

between groups were performed at breathing noise gain

of 3 using one way ANOVA. Significant findings were

post hoc tested using Bonferroni correction. Analysis

was performed with SigmaPlot 12 version 12.3 (Systat

Software, Inc. Chicago, IL).

2.2. Experiments

The simulation was used to verify the design and con-

struction of a prototype device.

Figure 1(a) shows a diagram of the bimorph actuator

and Figure 1(b) shows a diagram of the prototype FOT

device incorporating the bimorph actuator. An off-the-

shelf bimorph actuator (40-2010, American Piezo Cera-

mics, Mackeyville, PA) with length (L) of 60 mm, width

(w) of 20 mm and thickness (h) of 0.7 mm was used. The

piezoelectric bimorph was derived by an AC voltage

source using a function generator and a pre-amplifier and

a mesh screen was attached to actuator’s tip in order to

generate air oscillation at 6 Hz. The airflow was directed

into calibrated reference resistive loads of 5 and 15

cmH2O/L/s (Hans Rudolph Inc, KS, USA). Pressure of

airway opening was measured using a differential pres-

sure transducer (TD-05-AS, SCIREQ Inc., Montreal,

Canada) across the mesh screen that was attached to the

cantilever and flow was calculated by dividing the pres-

sure across pneumotachograph’s mesh by the screen’s re-

sistance (Rp = 0.4 cmH2O/L/s). Prior to the tests, trans-

Ta b le 1. Mechanical properties of subject populations, used in

Equations (4) and (5), and implemented in the model Figure

4(c) [8-11].

Property Healthy MaleCOPD Child

(8-year-old boy)

Asthmatic

Child

Rrs

(cmH2O/L/s) 2.35 10 6.86 9.5

Ers

(cmH2O/L) 33.3 60 82.84 100

Irs

(cmH2O/L/s2)0.0146 0.028 0.0092 0.0092

(a) (b)

Figure 1. (a) Structure of a bimorph actuator with parallel configuration. L, h and δ represent the length, thickness and displacement

respectively; (b) Schematic representation of the prototype device attached to the respiratory system model of different subjects for

simulations.

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071

1064

ducers were calibrated by applying steps of 0.5 cmH2O

using a manometer. Calibration coefficients were calcu-

lated based on the results. The breathing noise was gen-

erated using a ventilator pump (Bodine Electric Com-

pany, Chicago, ILL, USA). Measurements were repeated

for 6 times for each test. Pressure and Flow data were

sampled at 1000 Hz and the impedance was computed as

previously described [14]. The SNR was determined by

Fast-Fourier-Transforming (FFT) the entire duration of

data into the frequency domain, and calculating the ratio

of the magnitude of the FFT at the frequency of oscilla-

tion to the root mean squared average of noise in 1 Hz

bandwidth side bands adjacent to the oscillatory fre-

quency.

Section III describes the methods used to model the air

oscillometry device shown in Figure 1, including the

approach used to calculate the oscillation amplitude, and

to model the respiratory system and piezoelectric canti-

lever.

3. DESIGN AND MODELING

Unimorph, bimorph, and multimurph piezoelectric ac-

tuators are used in many different applications as a low

cost efficient means of converting electrical energy to

mechanical energy. Although the total potential energy

density remains the same for different multimorph ac-

tuators of the same geometry, the more layers incorpo-

rated into the multimorph, the lower the amplitude of the

required voltage is. Certain bimorph designs however

rely on a metallic spacer or vane separating the two rela-

tively thin piezoelectric plates. This type of bimorph ac-

tuator reduces the required electric field while maintain-

ing a reasonable displacement and actuator force at a frac-

tion of the cost of multimurph actuators. This is primar-

ily due to the less complicated manufacturing process.

Bimorphs are available in series and parallel configu-

rations. We chose a parallel configuration (Figure 1(a))

since they deflect the same amount as the series for half

of the series configuration’s applied voltage [5].

3.1. Calculations and Piezoelectric Model

Equation (1) presents the governing equation for oscil-

lating flow through a mesh screen affixed to the top of

the actuator that calculates the oscillation amplitude, δ, at

frequency, f:



 

 (1)

where 2π



 is the angular frequency, P is the mini-

mum pressure amplitude expected. Ra = 0.6 cmH2O/L/s

is the resistance of the actuating mesh screen and Ar =

πr2 is the area of the circular oscillating mesh of radius of

r = 2.45 cm.

Taking dielectric losses, structural damping and inertia

effects into account, a fixed-free cantilever bimorph ac-

tuator’s tip deflection (δ), as described in [5], can be pre-

dicted by Equation (2) as:

mC KV







  (2)

where Vin is the applied voltage across the thickness of

the bimorph, α is the electro-mechanical coupling, F is

the external load perpendicular to the cantilever’s tip, m

is the mass, Cd is the damping coefficient, and K is the

stiffness.

Figure 2 shows the electro-mechanical representa-

tion of the Butterworth-Van-Dyke (BVD) [15] model

of the piezoelectric actuator. The parallel capacitor, Cp

represents the electrical side of the model. The me-

chanical side represents a mass-spring-damperresonant

system.

Transforming Equation (2) into the frequency domain

with no load applied for the moment gives the electro-

mechanical transfer function (TFpiezo) of the piezo-elec-

tric actuator:



piezo

TFs msCs K



  (3)

m, K, Cd, and α where obtained by standard mechanical

experiments and using analytical formulas as follows

[16]. Briefly: To find m, we calculated and experimen-

tally measured the effective mass of the cantilever and

the mass on the tip needed to shift the bimorph resonance

down to 6 Hz. At resonance, K was determined from





 where r



is the resonant angular frequency.

The displacement versus frequency data was produced

from applying a chirp signal and recording the displace-

ment using a non-contact type laser displacement sensor

(AR 200, Acuity, Ortonville, MI). From the curve we

obtained the damping ratio









 where





the angular half-power bandwidth. The damping coeffi-

cient was computed as 2CKm



 

d. To find α,

this data was least mean squared fit to the transfer func-

tion (Equation (3)) with R2 = 0.97 (Ta ble 2). The move-

Figure 2. Electro-mechanical representation of Butterworth-

Van-Dyke (BVD) piezoelectric model.

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071 1065

Table 2. Properties of the piezoelectric cantileve; see text for

equations for fitting and experiment descriptions to determine

the properties.

Property Fitted Value Approach

α 2.6 N/V Fitting R2 = 0.97

Cp 190 nF Manufacturer

m 78 gr Experiment

Cd 0.53 Ns/m Experiment

K 110.8 N/m Experiment

ment amplification due to the distance between the tip of

the actuator and the center of the mesh screen is added to

the model by Lr





 where (L + r is the location of

the center of mesh.

3.2. Respiratory System Model and Impedance

To model the respiratory system impedance, we used the

standard single compartment model which is the equa-

tion of motion (Equation (4)). The resistance (RRs) was

modeled as a single Newtonian resistive tube. The elas-

tance (ERs) represents the stored elastic energy largely

from surface tension, but also from tissue stretching and

some gas compression while the intertance (IRs) repre-

sents the inertia of the moving gas (Figure 3) [17].

RsRs RS

PRVEVI V



(4)

where P is the airway pressure, V is volume, is flow

and is volume acceleration.





The respiratory system impedance, Zrs, is obtained by

transforming Equation (4) into the frequency-domain:



()

rs RsRS

ZjR jI



 (5)

This is well established model particularly well suited

for the healthy lung, and through altering Rrs, Ers and Irs,

the model can approximate the major changes in me-

chanics observed in common lung disease, although a

more accurate representation would include some in-

creasing frequency dependence of the parameters, par-

ticularly Rrs and Ers. For the purposes of design, we have

explored a range of mechanical parameters that describe

the gross changes commonly encountered using FOT

including COPD as well as a child [8-11] (Table 1). The

changes observed in asthma in an adult would be similar

to the increase in Rrs and Ers observed in a child.

4. RESULTS

4.1. Simulations

To verify the performance of the single frequency piezo-

electric forced oscillation device, simulations of the model

presented in Section III were performed (Figure 4(c)).

Figure 3. Single compartment model of the respiratory system

based on linear RLC circuit analogy: RRs is the resistance of the

tubing, ERs is the elastance of the tissue, and IRs is the inertance

of the flowing gases. Pawo is the pressure of airway opening and

Pms is the pressure created by muscle effort. is the breath-

ing flow/noise and



 is the oscillation flow generated by pi-

ezoelectric actuator.

The normal breathing of a representative subject with

frequency of 0.2 Hz containing the 6 Hz oscillations

shown in Figure 4(b) caused fluctuations of approxi-

mately 0.2 mm on actuator’s tip displacement (Figure

4(a)). This distortion increased with increasing noise

gain. As expected, when the breathing noise amplitude

was increased, the flow SNR dropped. The decrease was

more severe for the COPD and child simulation with

adulttidal volume (VT) than other subjects (p < 0.001)

and decreased to 28 dB at the simulated noise gain of 2.5,

while the SNR remained higher than 35 dB for noise

gains up to 3 for the male subject, the child with realistic

VT and the asthmatic child (Figure 5).

The variability in the results is due to the difference in

volunteer patients breathing patterns, however, the rank

ordering at every breathing noise gain stayed the same

amongst subjects.

Increasing the breathing noise up to three times the

normal values showed that the error in estimation of im-

pedance of the lung exceeded 10% in the COPD simula-

tion with breathing gain of 2 or more and in the child

simulation with adult tidal volume with breathing gain of

3.2 or more, however, the error in the child simulation

with realistic tidal volume, the asthmatic child and the

healthy male was less than 5% at any breathing noise le-

vel (Figure 6). For breathing noise gain of unity, the lar-

gest absolute error in estimating Rrs was 0.175 cmH2O/L/s

that represents 1.7% error, which belongs to the COPD

subject. The COPD subject also had the largest absolute

error in estimating Xrs of 0.045 cmH2O/L/s. It is not ap-

propriate to use % error representation for Xrs because its

values are close to zero at respiratory system resonant

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071

1066

(a) (b)

(c)

Figure 4. (a) Tip displacement affected by breathing noise gain of unity from simulations (b) Regenerated measured breathing flow

of a volunteer via a pneumotachograph with superimposed oscillations from simulations according to the same time scale as in (a). (c)

Schematic representation of the modeled system.

frequency. Also, Elastance and Inertance cannot be pre-

cisely distinguished due to the single frequency measure-

ment described here; therefore, the error in estimation of

Reactance is used (Figure 6).

Verifying the effect of leak due to air-gap on SNR and

the impedance estimation error showed that with total

resistance against the leak greater than 0.38 cmH2O/L/s

would give SNRs higher than 30 dB and errors lower

than 10% for the different simulated respiratory imped-

ances (Figure 7).

OPEN ACCESS

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071 1067

Figure 5. The effect of increasing breathing noise on measured

flow SNR ratio in different subject populations. Increasing the

breathing noise decreased the SNR. Error bars indicate SD for

all figures.

Figure 6. Magnitude estimation error of Zrs in percent for dif-

ferent subject populations at different noise levels. The estima-

tion error increased with increasing breathing noise. The insert

indicates Rrs and Xrs absolute estimation errors for unit breath-

ing noise gain.

Figure 7. SNR (a) and impedance estimation error (b) for different amounts of leak (mesh screen and airgap). As resistance against

the leak increases (leak decreases) the SNR increases and the estimation error decreases.

4.2. Nonlinearity

The respiratory system model used in this paper was a

linear single compartment model. However, there are non-

linearities that can arise such as form the pressure volume

curve, and the well-established Rohrer flow non-linearity

[18]. Here we explored the effects of the dependence of

the resistance of the airway on flow according to Rohrer

equation as follows (Equation (6)) that introduces a non-

linearity to the equation of motion (Equation (4)):

12u

RKKV





 (6)

where Ru is the nonlinear flow resistance in the upper

airway, K1 is the linear resistive term and K2 is the

additional resistive term depending on flow. To assess

the influence of this nonlinearity on the performance of

the device, the linear resistance of the respiratory system

was set equal to RRs so that with zero flow the respiratory

resistance would be the same as linear model parameter

presented in Tabl e 1. The nonlinear coefficient (K2) was

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071

1068

set to 0.2 cmH2O/L2/s2 as representative of an adult sub-

ject, 7.7 cmH2O/L2/s2 as representative of a normal child

subject and 11 cmH2O/L2/s2 as representative of an as-

thmatic child subject [12,19,20].

The effect of introducing this nonlinearity lowered the

SNR at all breathing gains with typically less than 3 dB

difference for all subjects except in the child with adult

VT where the SNR lowered for about 13 dB at normal

breathing level. The impedance estimation error increased

with breathing gain but remained less than 3% at normal

breathing amplitudes except for the child with adult VT.

The error in child with adult VT subject reached 10% at

breathing gain of 0.7 while COPD reached the maximum

recommended error at breathing gain of 1.7. The asth-

matic child, the normal child and the male subject reached

the maximum recommended error at breathing noise gain

of about 2 (Figure 8).

4.3. Experiments

To verify the feasibility of the single frequency piezo-

electric forced oscillation device in experiments, the pro-

totype device was tested on flow resistive test loads of 5

and 15 cmH2O/L/s as representatives of a normal and an

extremely high respiratory loads with and without gener-

ated breathing noise of unit gain using a ventilator pump.

Using the prototype device on test loads with no brea-

thing, the measurement error was less than 1% with SNR

of 39.9 dB for the 5 cmH2O/L/s test load and error less

than 5% with SNR of 30.1 dB for the 15 cmH2O/L/s test

load (Figure 9). Adding the simulated breathing of unit

gain through a ventilation pump had an apparent influ-

ence on the measurement error where it went up to 5%

with deteriorated SNR of 29 dB for the 5 cmH2O/L/s test

load while the measurement error on 15 cmH2O/L/s test

load went up to10% and its SNR dropped to 19.9 dB

(Figure 9).

5. DISCUSSION

5.1. Simulations

The aim of this paper was to design an efficient low po-

wer forced oscillation device that can meet the standards

proposed by Oostveen et al. [13] where they recommend

a maximum error of 10%. Simulation results presented in

the previous section showed that this criterion can be met

even with higher noise levels. The healthy male subject

and the asthmatic child were influenced similarly by the

increasing noise both in SNR drop and the increase in the

impedance estimation error. The impedance estimation

error in the simulated COPD and child (with high tidal

volume) subjects was higher than the healthy male sub-

ject (Figure 6). The same trend can be seen in Figure 5

where the SNR drop in simulated COPD and child im-

pedance was higher than the healthy male subject. To ve-

rify the mechanism for this, we calculated the transfer

function of the noise on the respiratory system of differ-

ent subjects (Appendix), taking into account the affect of

the breathing noise feedback on the oscillation dynamics

of the piezoelectric actuator.

Equation (7) presents the total noise transfer function

(TFNT):

Rs d

RsRS leak

leak RsRs

RIsR

smsCs K

RR Is













 











(7)

Figure 8. The comparison of the nonlinear and linear model in terms of SNR and impedance estimation error for increasing breathing

noise in a representative subject. Nonlinearity reduced the SNR and increased the error.

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071 1069

(a) (b)

Figure 9. (a) Schematic representation of the experimental test set up on the prototype device. The tests on the air-flow resistive test

loads were performed with and without ventilator generated breathing noise of unit gain; (b) Experimentally determined test load

measurements. The generated breathing noise increased the estimation error.

Figure 10 shows the magnitude of bode plot of Equa-

tion (7) around the frequency of oscillation for different

subjects. It can be noticed that the noise magnitude in

simulated COPD was the highest, the simulated child has

a lower peak while the average male impedance has the

lowest noise peak at 6 Hz. This is in agreement with the

results in Figures 5 and 6. It should be noted that the

peaks of the bode plot (Figure 9(b)) would shift in mag-

nitude in reality depending on the tidal breathing gain of

the specific subject. For example, as shown in Figure 5,

the simulated child subject with adult tidal breathing

exhibits a severe drop in SNR that resembles the COPD

simulation. This is because the mechanical impedance of

the child and COPD respiratory systems are both high in

value (Ta ble 1). However when the realistic amounts of

tidal volume is applied to the child’s respiratory system,

the noise decreases, and the SNR is maintained even high-

er than that of the healthy male simulation.

Producing oscillations of air through a moving mesh

screen includes the intrinsic problem of some of the re-

sistance originating from the mesh screen with the re-

mainder arising from the gap between the mesh and the

housing tube. Considering the impedance using an elec-

trical analog model, the resistance of air-gap is in parallel

with mesh resistance and thus the total equivalent resis-

tance is smaller than the mesh resistance. In this work the

equivalent resistance of the mesh screen and air-gap was

defined to be 0.6 cmH2O/L/s (Figure 7). An interesting

conclusion from Figure 7 is that if the resistance against

the leak due to air-gaps in the device is maximum (open

circuit), the mesh screen can be replaced with one with

resistance as low as 0.38 cmH2O/L/s while maintaining

the impedance estimation error less than 10% and the

SNR higher than 30 dB.

Figure 10. Magnitude of the bode diagram of noise transfer

function and the noise magnitude vs impedance at 6 Hz. In-

creasing the impedance increased the noise.

Simulating the effects of nonlinearity on different simu-

lated subjects had the most substantial effect on the im-

pedance estimation error of the child with adult VT. The

effects on the male, the child with realistic VT and the

asthmatic child subject were also substantial while the

effect on the COPD was the smallest (Figure 8). The

small effects of the nonlinearity in the COPD subject

were because the nonlinear term was small relative to the

linear resistance, while in the healthy male the linear re-

sistance is small relative to the nonlinear term. However,

while nonlinearity led to increases in errors, the errors

were only unacceptable for the healthy male, the realistic

child and the asthmatic child subjects at larger flows than

normal. At normal flow levels and at the normal tidal

volume for the child subject, the error was acceptable in

all subjects.

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071

1070

5.2. Experiments

The tests on the prototype device without breathing noise

resulted in very low errors and high SNRs. As expected,

after adding the breathing noise the SNR dropped and

errors increased. However in the case of the 15 cmH2O/L/s

test load which is higher than would be anticipated for

adult subjects, but possible for children with airway ob-

struction [13] the error reached the maximum recom-

mended value (Figure 9(b)). Children fortunately have

lower breathing noise, and thus this design is likely suit-

able for a large range of subjects. These are for normal or

nearly normal tidal volumes. However if measurements

are to be made under larger ventilation conditions such

as with exercise, the reduction in SNR would begin to

cause larger errors in estimating impedance. If this is un-

acceptable then larger oscillatory flows would be recom-

mended using one or a combination of following techni-

ques:

 Movement amplification techniques including me-

chanical lever and/or using longer customized piezo-

electric multimurphs and/or using multiple actuators

at the same time for increasing the applicable force.

 Using an actuator with higher surface area to generate

higher pressure oscillations.

 Reducing the amount of leak due to air-gaps.

6. CONCLUSION

In this paper, for the first time, an oscillometry device

was designed based on optimization in a simulation that

included the device characteristics together with changes

in subject impedance including perturbations from brea-

thing. Although breathing noise reduced the SNR, the per-

formance remained acceptable and demonstrated a useful

design approach that led to the development of a feasible

accurate lightweight portable single frequency FOT de-

vice. Reducing the leak and improving the measurement

accuracy of the transducers as well as modifications in

the oscillating mesh and the actuator can improve the

results from the prototype device.

7. ACKNOWLEDGEMENTS

Hamed Hanafi and Lucas Posada were supported by the NSERC

CREATE program at Dalhousie University, S. A. Bhatawadekar was

supported by the NSERC.

The authors thank Guy Drapeau at Thorasys Medical Systems for his

great feedback especially for experiments. They also thank Andre Be-

zanson for his help with piezoelectric actuators.

REFERENCES

[1] DuBois, A.B., Brody, A.W., Lewis, D.H., Burgess Jr.,

B.F. (1956) Oscillation mechanics of lungs and chest in

man. Journal of Applied Physiology, 8, 587-594.

[2] Dellacà, R.L., Pompilio, P.P., Walker, P.P., Duffy, N., Pe-

dotti, A. and Calverley, P.M. (2009) Effect of bronchodi-

lation on expiratory flow limitation and resting lung me-

chanics in COPD. European Respiratory Journal, 33,

1329-1337.

http://dx.doi.org/10.1183/09031936.00139608

[3] Morita, T. (2003) Miniature piezoelectric motors. Sensors

and Actuators A, 103, 291-300.

http://dx.doi.org/10.1016/S0924-4247(02)00405-3

[4] Uchino, K., Kato, K. and Tohda, M. (1988) Ultrasonic li-

near motors using a multilayered piezoelectric actuator.

Ferroelectrics, 87, 331-334.

http://dx.doi.org/10.1080/00150198808201395

[5] Wang, Q.-M., Du, X.-H., Xu, B.M. and Eric Cross, L.

(1999) Electromechanical coupling and output efficien-

cy of piezoelectric bending actuators. IEEE Transactions

on Ultrasonics, Ferroelectrics, and Frequency Control,

46, 638-646. http://dx.doi.org/10.1109/58.764850

[6] Cheng, H.-M., Ewe, M.T., Bashir, R. and Chiu, G.T.-C.

(2001) Modeling and control of piezoelectric cantilever

beam micro-mirror and micro-laser arrays to reduce im-

age banding in electro photographic processes. Journal of

Micromechanics and Microengineering, 11487-11498.

[7] Goli, J., Smits, J.G. and Ballato, A. (1995) Dynamic bi-

morph matrix of end-loaded bimorphs. IEEE Internatio-

nal Frequency Control Symposium, 794-797.

http://dx.doi.org/10.1109/FREQ.1995.484086

[8] Brown, N., Xuan, W., Salome, C., Berend, N., Hunter, M.,

Musk, A., James, A. and King, G. (2010) Reference equ-

ations for respiratory system resistance and reactance in

adults. Respiratory Physiology & Neurobiology, 172,

162-168. http://dx.doi.org/10.1016/j.resp.2010.05.013

[9] Navajas, D., Alcaraz, J., Peslin, R., Roca, J. and Farre, R.

(2000) Evaluation of a method for assessing respiratory

mechanics during noninvasive ventilation. European Re-

spiratory Journal, 16, 704-709.

http://dx.doi.org/10.1034/j.1399-3003.2000.16d23.x

[10] Van Den Bergh, A. and Kerrebijn, K. (1985) Forced os-

cillation technique. Reference values for resistance and

reactance over a frequency spectrum of 2 - 26 Hz in heal-

thy children aged 2.3 - 12.5 years. Bulletin of European

Physiopathological Respiration, 21, 171-178.

[11] Lall, C.A., Cheng, N., Hernandez, P., Pianosi, P.T., Dali,

Z., Abouzied, A. and Maksym, G.N. (2007) Airway resis-

tance variability and response to bronchodilator in chil-

dren with asthma. European Respiratory Journal, 30,

260-268. http://dx.doi.org/10.1183/09031936.00064006

[12] Barbini, P., Cevenini, G. and Avanzolni, G. (2003) Non-

linear mechanisms determining expiratory flow limitation

in mechanical ventilation: A model-based interpretation.

Annals of Biomedical Engineering, 31, 908-916.

http://dx.doi.org/10.1114/1.1590665

[13] Oostveen, E., MacLeod, D., Lorino, H., Farre, R., Hantos,

Z., Desager, K. and Marchal, F. (2003) On behalf of the

ERS task force on respiratory impedance measurements:

“The forced oscillation technique in clinical practice: Me-

thodology, recommendations and future developments.”

European Respiratory Journal, 22, 1026-1041.

http://dx.doi.org/10.1183/09031936.03.00089403

H. H. Alamdari et al. / J. Biomedical Science and Engineering 6 (2013) 1062-1071

1071

[14] Bhatawadekar, S.A., Leary, D., Chen, Y., Ohishi, J., Her-

nandez, P., Brown, T., McParland, C. and Maksym, G.N.

(2013) A study of artifacts and their removal during for-

ced oscillation of the respiratory system. Annals of Bio-

medical Engineering, Epub ahead of print.

http://dx.doi.org/10.1007/s10439-012-0735-9

OPEN ACCESS

[15] Van Dyke, K.S. (1928) The piezo-electric Resonator and

its equivalent network. Proceedings of the IRE, 16, 742.

[16] Inman, D. (2001) Engineering vibration. Prentice Hall,

Upper Saddle River.

[17] Jason H.T.B. (2009) Lung mechanics: An inverse model-

ing approach. Cambridge University Press, Cambridge.

http://dx.doi.org/10.1017/CBO9780511627156

[18] Rohrer, R. (1915) Der stroemungswiderstand in den men-

schlichenatemwegen und der einfluss der unregelmaessi-

genverzweigung des bronchialsystems auf den atmungs-

verlauf in verschiedenenlungenbezirken. American Jour-

nal of Physiology, 162, 225-229.

http://dx.doi.org/10.1007/BF01681259

[19] Bouhuys, A. and Jonson, B. (1967) Alveolar pressure, air

flow rate, and lung inflation in man. Journal of Applied

Physiology, 22, 1086-1100.

[20] Schweitzer, C., Chone, C. and Marchal, F. (2003) Influ-

ence of data filtering on reliability of respiratory imped-

ance and derived parameters in children. Pediatric Pul-

monology, 36, 502-508.

http://dx.doi.org/10.1002/ppul.10359

APPENDIX influence on oscillations of piezo actuator. Transforming

Equation (4) into s-domain we have the transfer function

of the respiratory system as below:

The aim of this appendix is to calculate the total noise

transfer function that appears in output pressure includ-

ing the effect of breathing noise on the screen mesh. Rs

Rs RsRS

TFRI s



 (A.3)

From Figure 4(c) we can write:



inPP bdP

VKxxxC





 





 x



(A.1) Therefore we can write (Figure 4(c)):

awo

bb NORs aw

VVTFTFP



 







(A.4)

If we assume Vin = 0 and knowing that b





 and

Rearranging and using Equations (A.2) and (A.3), giv-

es the total noise transfer function (TFNT) of:

VxAr



 , with rearrangement we have:

TF VmsCs K













 (A.2)



Rs leakNO

awo

leak Rsb

TF RTF

TF RTFV



 

 (A.5)

Where TFNO is the transfer function of feedback noise