JOURNAL OF AEROSOL MEDICINEVolume 19, Number 4, 2006© Mary Ann Liebert, Inc.Pp. 533–542

Administration of Aerosolized Drugs to Infants by aHood: A Three-Dimensional Numerical Study

TAL SHAKKED, M.S.,1 DAVID M. BRODAY, Ph.D.,2 DAVID KATOSHEVSKI, Ph.D.,1and ISRAEL AMIRAV, M.D.3

ABSTRACT

Using a hood for aerosol delivery to infants was found to be effective and user-friendly com-pared to the commonly used face mask. The currently available hood design has an evengreater potential in terms of efficiency, and a numerical simulation can serve as a tool for itsoptimization. The present study describes the development and utilization of a numericalsimulation for studying the transport and fate of the aerosol particles and the carrier gas withina three-dimensional realistic representation of the hood and the infant’s head. The study fur-ther incorporates realistic breathing patterns, with a longer expiration phase than an inspira-tion phase. Both nose and mouth breathing are simulated. While the base case assumes thatthe funnel that delivers the aerosol within the hood is perpendicular to the infant’s face, morerealistic scenarios include a funnel that is slanted with respect to the infant face, the infant’shead taking a general position with respect to the funnel, and the funnel and the head beingboth tilted. A good agreement is found between computation and experimental results. Asexpected, the most efficient drug delivery, 18%, is achieved when the funnel is normal to theinfant’s face. The quantitative evaluation of different scenarios presented in this work in-creases the knowledge of physicians, nurses, and parents regarding the efficacy of the treat-ment, in terms of the actual amount of drug inhaled under various modes of function of thedevice.

Key words: aerosol therapy, breathing function, nebulizer

533

INTRODUCTION

AEROSOL DELIVERY to infants using a face maskis known to have several disadvantages in

terms of patient’s tolerance and efficient handlingby non-professionals. A major drawback is thedifficulty of achieving a good mask–face sealwhen the infant is screaming and crying.(1,2)

Moreover, nebulizer treatments can take 10–15min, much longer than most infants can toleratewhen using a mask. The infant’s impatience fur-ther reduces the efficiency of drug delivery to thelungs.(3,4) Recently, aerosol therapy to wheezy in-fants using a hood interface has been reported tobe as efficient as using a mask.(5) As expected, thehood was preferred by parents and better toler-

1Department of Biotechnology and Environmental Engineering, Institute for Applied Biosciences, Ben-Gurion Uni-versity of the Negev, Beer-Sheva, Israel.

2Faculty of Civil and Environmental Engineering, Technion–Israel Institute of Technology, Haifa, Israel.3Pediatric Department, Sieff Hospital, Safed, Israel.

ated by the infants. This led to the developmentof a hood-shaped device (Fig. 1).

The original device was the subject of a study(6)

in which an axisymmetric geometry was consid-ered. In the present study, a more realistic andrepresentative configuration is considered. Thisstudy aims at elucidating issues that could not beaddressed using the former two-dimensional (2-D) simulation. Namely, while the focus of the for-mer work was on the general characteristics ofthe hood and the flow within it, this work focuseson features that can be addressed only when us-ing a true three-dimensional (3-D) geometry,such as the actual kind of nasal breathing prac-ticed by infants 80% of the time. Specifically, thiswork considers a detailed 3-D configuration ofthe hood and the infant’s head, including thechin, shoulders, mouth, nose, and nostrils, withall dimensions based on data from an actual 5-month-old infant. This model enables us to in-vestigate drug administration while the infant isbreathing through the nose or the mouth, as wellas when the funnel and the head are not tilted co-linearly. Moreover, a full transient solution hasbeen calculated, which is necessary when dealingwith realistic time-varying breathing functionslike those implemented in this study at the open-ings of respiratory system. Normally, such abreathing pattern includes regular tidal breathingwith inspiratory duty cycle (IDC)—the fraction ofinspiration time to the breathing period—of 0.4.(7)

METHODS

Nebulizer hood

The nebulizer hood (Baby’s Breath AdvancedInhalation Technologies, Israel) was describedpreviously.(6) In short, the hood system in-cludes a funnel (effective volume of 238 cm3),a pneumatic nebulizer, a hemispherical andflexible plastic cape that encloses the infant’shead, and four folding legs. The funnel delivers the aerosolized drug to the infant. Itcan be manually adjusted for better perfor-mance by moving it up and down and by tilt-ing it in any direction at a wide range of angles(Fig. 1). The end of the funnel is narrower inorder to direct the aerosol towards the open-ings of the infant’s respiratory system. The neb-ulizer is attached at the top of the funnel withan adapter and is driven by a small-sized com-pressor that operates at a flow rate of 7 L/minand provides a mass flux of 0.25 g/min. Drugdroplets of about 5 �m in diameter are pro-duced by the nebulizer.

Governing equations

A commercial computational fluid dynamics(CFD) software (FLUENT 6.1, Fluent Inc.),which is based on the finite volume method,was used to solve the governing equations sub-ject to time varying boundary conditions.Specifically, the flow field of the air and the dis-crete phase were taken to be non-steady, the in-jection of aerosol was assumed continuousthroughout the cycle, and realistic breathingpatterns were implemented at the openings ofthe respiratory system (the mouth and the nos-trils). The flow field was obtained by solvingthe mass and momentum conservation equa-tions. The continuity equation for the continu-ous phase (air) is:

� �� �u�, (1)

where u is the velocity vector and � is the air den-sity. The above equation is written in its form forthe sake of generality. However, in the numeri-cal procedure the continuous phase is consideredincompressible.

The momentum equation for the air is

�� � u� � �u�� � ��� � �p � �g�, (2)�u���t

d��dt

SHAKKED ET AL.534

FIG. 1. The nebulizer hood inhaler. The arrows showpossible modifications of the funnel.

where p is the pressure, g� is the gravitational ac-celeration, �� is the stress tensor,

�� � ��(�u� � �u���), (3)

and � is the dynamic viscosity of air. Since Gr/Re2

1, where Gr is the Grashof number and Re isthe flow Reynolds number, natural convectiondue to non isothermal conditions of the air canbe neglected(6) (i.e., apart from a narrow layernear the mouth and the nostrils, the flow field inthe hood is unaffected by temperature disparitydue to an efficient convective mixing).

The dispersed phase is simulated using a La-grangian frame of reference. This phase consistsof spherical aerosol particles (droplets) dispersedin the continuous phase. Particle trajectories arecalculated by accounting for the total force actingon the particles. Conservation of momentum forthe particles reads,

� FD(u� � u�p) � g�(�p � �)/�p (4)

where the index p denotes particle properties. Inour case �p �, hence the buoyancy term isnegligible. The drag term is given by

FD � (5)

with u� and u�p being the velocity vectors of thefluid phase and the dispersed phase, respectively,�p is the particle density, and dp is the particle di-ameter. The particle Reynolds number Re is de-fined as

Re � (6)

and the drag coefficient, CD, is

CD � a1 � � (7)

where the a’s are empirical constants.(10) Integra-tion of Eq. (4) yields the particle velocity. The par-ticle position is calculated then by a second inte-gration,

� up,i (8)

It is important to note that, while the effect ofair motion on particle dynamics is accounted forby Eq. (4), the effect of particle motion on the airflow is not included. This is often the case in aerosol-related studies,(11) although, strictly

dxi�dt

a3�Re

2a2�Re

�dp�u�p � u����

�

CDRe�24

18���pdp

2

du�p�dt

speaking, it applies only to dilute aerosols.(12,13)

Since the particle mass and volume fractions inthe following simulations are of the order of 10�2

and 10�3, respectively, the effect of particle mo-tion on the airflow can be neglected.

Computational domain and mesh generation

Our goal was to investigate realistic scenariosthat parents and hospital staff may face on dailybasis. Therefore, we chose the most commoncases: the funnel being perpendicular to the in-fant’s face, the funnel being tilted with respect tothe head, the head being tilted with respect to thefunnel, and the funnel and the head being bothin a general inclination. A realistic representationof the infant’s face and head required that the sur-faces will be generated by many curved lines andplanes, forming an object with a relatively com-plex geometry. The dimensions of these elementswere measured in a 5-month-old infant. A threedimensional (3-D) geometrical computational do-main was built based on these dimensions foreach of the four funnel-head configurations men-tioned above. The computational domain wascreated using the GAMBIT package, which buildsand meshes models for computational fluid dy-namics (CFD) and other numerical studies.(8) Dueto the complex geometry, in some of the cases cer-tain nodes were placed individually and the meshdensity between the funnel and the head was re-fined. The mesh (about 60,000 cells) containedprimarily of tetrahedral elements, but hexahe-dral, pyramidal, and wedge elements were alsoused. The mesh density has been increased untilgrid independent solution was achieved. Modelelements include the head, the mouth (an ellipsewith area of 5 cm2), the nostrils (circles with areaof 28 mm2), and the funnel (Fig. 2). In the basecase, the funnel’s end is 1.5 cm above the infant’snose (2 cm above the nostrils). It is assumed thatthe infant is laying on a flat surface representingthe bed.

Breathing functions

Common infant breathing patterns are mod-eled via breathing functions derived using thePOLYMATH software package. The breathingfunctions represents regular tidal breathing withinspiratory duty cycle of 0.4, based on a sinu-soidal waveform.(9) Regular tidal breathing ischaracterized by a 80-cm3 tidal volume and 2-sec

3-D SIMULATION OF HOOD INHALER 535

breathing period. The breathing function fornasal and mouth breathing is shown in Figure 3.

Boundary conditions

For the continuous phase, velocity boundaryconditions were specified at the entrance to thecomputational domain and at the openings ofthe respiratory system. The latter were specifiedusing the appropriate breathing functions de-picted in Figure 3, namely, tidal breathing withinspiratory duty cycle of 0.4 based on a sinu-soidal waveform. The flow rate through the fun-nel was set to 7 L/min. No-slip condition wasspecified at the walls and atmospheric pressurewas specified at the gap between the hood andsurface on which the infant is laying (the pres-sure condition was introduced as zero gaugepressure). During nasal breathing, velocitycomponent in the negative z-direction indicatesinspiration whereas a positive z-component in-dicates expiration. During mouth breathing, avelocity component in the negative y-direction

indicates inspiration whereas a positive y-ve-locity component indicates expiration.

For the dispersed phase, particle velocity (samevelocity as the continuous phase),* size and cross-sectional distribution at the entrance to the fun-nel were given. On all solid boundaries a“trapped” condition was applied, namely parti-cle is trapped when its center reaches within adistance of its radius from the wall. At the cir-cumferential slit and at the openings of the res-piratory system an “escape” condition is applied,implying that a particle passing these boundariesleaves the computation domain. and the trajec-tory calculation is terminated.

Numerical approach

In order to model the transient problem prop-erly, the time step for the transient simulationwas set to 0.05 sec. Trajectories of the discretephase were calculated by stepwise integrationover discrete time steps, with the integration timestep set by FLUENT to obtain a minimum error.Integration of equation 4 with the trapezoidalscheme yields the velocity of the particle at eachpoint along its trajectory. Care was taken that theparticle number would indeed not interfere withthe flow field.

Operating conditions

Four typical configurations of the funnel rela-tive to the infant’s head were analyzed: (1) thefunnel being perpendicular to the infant’s face,(2) the head being tilted relative to the funnel, (3)the funnel being tilted in several angles relativelyto the infant’s head, and (4) both the funnel and head are being non-co-linear. Accordingly, simu-lation results are classified according to the spa-tial relationships between the funnel and thebaby’s head. Within each of these general sce-narios several cases that correspond to differentbreathing parameters were studied. These casesreflect inter-subject physiological/behavioralvariations among the patients. In the base sce-nario, the funnel is perpendicular to the infant’sface. The cases studied under this scenario in-clude nasal tidal breathing with inspiratory dutycycle of 0.4, mouth tidal breathing with inspira-

SHAKKED ET AL.536

FIG. 2. A three-dimensional representation of the neb-ulizer hood inhaler, and the infant’s head and shoulders.A finer model was implemented and used in the compu-tational fluid dynamics (CFD) simulations.

*A computational software that allows the user to ap-ply effective numerical analysis techniques during inter-active problem solving on personal computers.

tory duty cycle of 0.4, and nasal tidal breathingwith inspiratory duty cycle of 0.4 when repre-sentation of the lips is included. The second sce-nario considers nasal breathing when the funnelis tilted relative to the infant’s face. Sub casesstudied include a funnel tilted in the x direction(sideways), a funnel tilted in the z direction (inthe sagital plane), and a funnel tilted at a generalinclination. The third scenario is when the headis tilted relative to the funnel. Here we considerone case: nasal breathing with the head tiltedsideways (in the x direction). The last scenarioconsiders nasal breathing when the head and thefunnel are in a general non-collinear orientation.The two cases studied are (1) the head and thefunnel are tilted sideways (in the x direction) and(2) the head is tilted sideways (in the x direction)while the funnel takes a general orientation.

RESULTS

When the funnel is vertical and normal to theinfant’s face, the computational geometry is sym-metrical (Fig. 4a). Here, like in the 2-D axisym-metric case,(6) the air accelerates as it moves to-ward the funnel’s narrow exit and the velocityfield within the funnel and at short distancesaway from it is not affected by the infant’s breath-ing pattern. For this base case configuration,model results revealed that after six completebreathing cycles with continuous injection ofaerosol particles at the entrance to the computa-tion domain (approximately 5000 particles pertime step), 18% of the particles were inhaledthrough the nostrils of the infant. The velocityvectors at the mid-time of inspiration and expi-ration phases are depicted in Figure 5a,b. It is

3-D SIMULATION OF HOOD INHALER 537

FIG. 3. Representation of a nasal and mouth breathing function of IDC � 0.4.

shown that during the inspiration phase less thanhalf of the air-flux exiting the funnel is flowingthrough the nostrils, and at expiration there is noflow through the nostrils at all. Since the lips per-turb from the face to some degree, they may havea local effect on the flow field near the nostrils.In order to test whether this has any effect onaerosolized drug administration, we imple-mented a computational geometry with lips (Fig.6). The dimensions and shape of the lips weretaken from the same 5-month-old infant used for creating the computational geometry. Theamount of aerosol entering the respiratory sys-tem through the nostrils was identical to that cal-culated under the no-lips case, suggesting that thelips do not affect the aerosol inhalation throughthe nostrils. In fact, the funnel was positioned justabove the infant’s nose even when breathingthrough the mouth was considered. This breath-ing mode is not common among infants but mayreflect special conditions, such a rhinitis. Undermouth breathing, only 7% of the aerosol intro-duced at the entrance to the funnel penetrates themouth. The velocity vectors at the mid-time of in-

spiration and expiration phases are depicted inFigure 5c,d, which shows that most of the flowfrom the funnel does not reach the mouth duringinspiration, and at the expiration phase there isno flow through the mouth at all.

Under normal operating conditions, we expectthat the funnel, the head, or both will be inclined.Therefore, we implemented configurations whichfrequently occur in realistic scenarios. When thefunnel is tilted at 10° sideways (along the x-axis)and the head is at its base configuration (Fig. 4b),simulation of nasal breathing with IDC (inspira-tory duty cycle) of 0.4 revealed a decrease in theamount of aerosol particles penetrating throughthe nostrils (3.5%). This is supported by the ve-locity vectors at the inspiration phase shown inFigure 5e, where flux is not entering the respira-tory system. A considerable amount of drug istherefore lost, since most of the air that emanatesfrom the funnel does not approach the nostrilsbut is rather drifted away. When the funnel istilted at 10° along the z-axis (in the longitudinaldirection) and the head is at its base configura-tion (Fig. 4c), the amount of drug delivered to the

SHAKKED ET AL.538

FIG. 4. Different configurations of the funnel and the infant’s head. (a) Base case—the funnel is normal to the in-fant’s face. (b) The funnel is tilted sideways (along the x axis). (c) The funnel is tilted in the longitudinal direction(along the z axis). (d) The funnel takes a general inclination with respect to the vertical. (e) The head is tilted side-ways (in the x direction), while the funnel is vertical. (f) The head and funnel are tilted sideways towards each other.(g) The head is tilted sideways (along the x axis), and the funnel is arbitrarily inclined to the vertical.

nostrils is 4.5%. The velocity vectors of the inspi-ration phase are depicted in Figure 5f. Again, aconsiderable amount of drug is lost, since mostof the air that emanates from the funnel does notapproach the nostrils but rather drifts away. Thesharp decrease in drug delivery in these two lastconfigurations relative to the base case is due tothe non-axisymmetric nature of the problem. In

the more general case, when the funnel is posi-tioned at an angle of 10° to the x-z plane (Fig. 4d)only about 5.5% of the particles reach the nostrils.The velocity vectors at the inspiration phase aredepicted in Figure 5g. This result shows the non-additive (i.e., non-linear) nature of the inhalabil-ity with respect to geometrical variations of ele-ments that build the scenario.

3-D SIMULATION OF HOOD INHALER 539

FIG. 5. Vectors of velocity magnitude (m/sec) for different configurations. (a) Base case—the funnel is normal tothe infant’s face, mid-time of inspiration, nose breathing. (b) Base case—the mid-time of expiration, nose breathing.(c) Base case—the mid-time of inspiration, mouth breathing. (d) Base case—the mid-time of expiration, mouth breath-ing. (e) The funnel is tilted sideways (along the x axis), the mid-time of inspiration, nose breathing. (f) The funnel istilted in the longitudinal direction (along the z axis), the mid-time of inspiration, nose breathing. (g) The funnel takesa general inclination to the vertical, the mid-time of inspiration, nose breathing. (h) The head is tilted sideways (inthe x direction), while the funnel is vertical, middle of inspiration, nose breathing. (i) The head and funnel are tiltedsideways towards each other, middle of inspiration, nose breathing. (j) The head is tilted sideways (along the x axis),and the funnel is arbitrarily inclined to the vertical, middle of inspiration, nose breathing.

When the head is tilted sideways (in the x di-rection) and the funnel is kept vertical (Fig. 4e),only 6% of the particles introduced to the systempenetrates the nostrils. The velocity vectors at theinspiration phase are shown in Figure 5h. Com-pared to the case when the funnel is tilted side-ways and the head is at its base configuration,this result clearly shows that inhalability is non-commutative with respect to variations in thescene. In reality, oftentimes both the funnel andthe head are non-co-linear. This scenario was ex-amined with both the head and the funnel tiltedat 15° sideways towards the same direction (Fig.4f). About 14.5% of the drug particles were in-haled through the nostrils. The velocity vectors atinspiration phase are depicted in Figure 5i. Com-pared to the other scenarios (excluding the basecase scenario), when the funnel is inclined in thesame direction as the mouth and the funnel exitis just above the nostrils, the efficiency of drugdelivery is relatively high. Another realistic con-figuration is when the head is tilted sideways

while the funnel is inclined randomly with re-spect to the x-z plan. One possible configuration,simulated here, is when the head is tilted side-ways at 15° and the funnel makes an angle of 15°with the normal x-z plane (Fig. 4g). Nasal breath-ing under such conditions results in inhalabilityof about 1.5%. The velocity vectors at the inspi-ration phase are depicted in Figure 5j. A summaryof all the scenarios is presented in Table 1.

Finally, a combined mouth and nose breathingwas studied, with 80% of the air inhaled throughthe nose. Here we studied how long it takes forthe flow field to stabilize after switching betweennose and mouth breathing. Simulation resultssuggest that the flow field stabilizes during thefirst breath after switching from mouth to nose,or from nose to mouth breathing. For 80/20%nose/mouth breathing, the amount of aerosol en-tering into the respiratory system is 15.8%.

An IDC of 0.4 is a general characteristic of in-fants’ breathing. However, due to normal vari-ability, different infants will have slightly differ-ent IDCs. To study the effect of that variability ondrug delivery, different IDC’s were simulated(IDC � 0.35, 0.45, and 0.5). A linear relationshipbetween the IDC and the percentage of aerosoldelivered to the infants nostrils was found (Table2). This behavior is expected since, as the inspi-ration period increases, the amount of inhaledaerosols increases comparably.

The effect of submicron particles was also ex-amined. A simulation with 0.5-�m particles wasexecuted, and it was found that the change in per-centage of particles arriving to the inlet of the res-piratory system is negligible compared with the5-�m case. This coincides with the same conclu-sion reported in the author’s previous work.(6)

SHAKKED ET AL.540

TABLE 1. RESULTS OF DIFFERENT SCENARIOS

Percentageof inhaled

Case Description aerosols

1 Base case: vertical funnel, IDC � 0.4, nasal breathing 18.%2 Base case � lips 18.%3 Vertical funnel, IDC � 0.4, mouth breathing 7.%4 Upright head, funnel tilted sideways at 10° 3.5%5 Upright head, funnel tilted longitudinally at 10° 4.5%6 Upright head, funnel tilted at 10° in a general direction 5.5%7 Head tilted sideways, vertical funnel 6.%8 Head and funnel tilted at 15° sideways 14.5%9 Head tilted at 15° sideways and funnel tilted at 15° in a 1.5%

general direction

IDC, inspiratory duty cycle.

FIG. 6. Computational geometry, including lips.

Model validation

Experimental data was used to validate themodel. The experiment was conducted by Baby’sBreath Advanced Inhalation Technologies, Israel.The experimental setup resembled the base casescenario implemented in the model in terms ofthe operating conditions and the configuration ofthe nebulizer hood (Fig. 7). The experiment hasbeen done with a doll with general dimensionsof an infant, in which a filter was placed few cen-timeters behind the nostrils. The nebulizer hoodwas set to work at the same operating conditionsas those implemented in the simulations, and anartificial breathing simulator imitated real infantsbreathing parameters: a tidal volume of 75 cm3,breath length of 2 sec, and an IDC of 0.4 (0.8 secinspiration and 1.2 sec expiration). The end of thefunnel was located at the same distance from thenose as in the model. The ratio of the amount ofthe radioactive labeled particles that were col-

lected on the filter to the amount introduced intothe nebulizer was determined. The averageamount of radioactivity entering to the respira-tory system through the nostrils during 10 re-peating sessions was 17.82 � 2.04%. This result isin excellent agreement with the model result(18 � 2%), suggesting that the numerical modelis reasonably accurate and can indeed be used tostudy the efficiency of the hood nebulizer undera wide range of operating conditions.

DISCUSSION

A 3-D model for studying aerosolized drug de-livery to infants using the nebulizer hood has beendeveloped. The advantage of studying this systemusing a CFD model is the ability to examine dif-ferent realistic scenarios while avoiding the con-founding factors that are unavoidable in clinicalexperiments. This study, together with our formerstudy,(6) extend our understanding of the hoodand its operation. As expected, the most efficientway to deliver aerosol particles to the openings ofthe respiratory system using the nebulizer hood iswith the funnel normal to an infant’s face. Anyother configuration reduces the hood efficiency. Inparticular, a considerable decrease in the efficiencyoccurs when the funnel, the head, or both are tilted.In such cases, the infant’s breathing is not intenseenough to overcome the inertia of the aerosol thatemanates from the funnel. A major part of it (andits active agents) is therefore directed away from

3-D SIMULATION OF HOOD INHALER 541

TABLE 2. EFFECT OF DIFFERENT IDCS DURING NASAL

BREATHING AT THE BASE CASE CONFIGURATION ON THE

PERCENTAGE OF INHALED AEROSOLS

Percentageof inhaled

Case IDC aerosols

1 0.35 15%2 0.45 21%3 0.50 24%

IDC, inspiratory duty cycle.

FIG. 7. The experimental configuration.

the openings of the respiratory system and is lost.When both the head and the funnel are tilted, aninclination of the funnel toward the face is obvi-ously advantageous and succeeds in compensat-ing somewhat for the non collinear configuration.However, when the funnel is directed away fromthe infant’s face the overall system efficiency is re-duced significantly. These results suggest that cor-rect operation of the hood, more than a variationof its design, is the key parameter for success inadministering aerosolized therapeutics using thenebulizer hood. Thus, the caregiver should be ad-vised to verify that during nebulizer hood treat-ments the funnel should always be positionedsuch that it is directed towards the head. Ignoringthis rule will hamper the efficacy of the treatmentconsiderably and require extended treatment du-rations to achieve the same results (dose).

Ample cases in different fields proved that crit-ical engineering-oriented CFD study is capable ofimproving product design and performance. Thiswork shows how careful modeling can improvethe modus operandi of delivery systems ofaerosolized medications by providing guidelinesfor setting the user interface for optimal perfor-mance.

ACKNOWLEDGMENTS

We wish to thank Baby’s Breath Advanced In-halation Technologies, Israel, for their assistancein carrying out this research

REFERENCES

1. Everard, M.L., A.R. Clark, and A.D. Milner. 1992.Drug delivery from jet nebulizers. Arch. Dis. Child.67:586–591.

2. Amirav, I., and M.T. Newhouse. 2001. Aerosol ther-apy with valved holding chambers in young children:importance of the facemask seal. Pediatrics 108:389–394.

3. Tal, A., H. Golan, N. Grauer, et al. 1996. Depositionpattern of radiolabeled salbutamol inhaled from me-

tered-dose inhaled by means of a spacer with maskin young children with airway obstruction. J. Pediatr.128:479–484.

4. Murakami, G., T. Lgarashi, Y. Adachi, et al. 1990. Mea-surements of bronchial hyperreativity in infants andpreschool children using a new method. Ann. Allergy64:383–387.

5. Amirav, I., I. Balanov, M. Gorenberg, et al. 2003. Neb-ulizer hood compared to mask in wheezy infants:aerosol therapy without tears. Arch. Dis. Child.88:719–723.

6. Shakked, T., D. Katoshevski, D.M. Broday, et al. 2005.Numerical simulation of air flow and medical-aerosoldistribution in an innovative nebulizer hood. J. AerosolMed. 18:207:217.

7. Levitzky, M.G. 1995. Pulmonary Physiology. Mc-Graw-Hill, London.

8. GAMBIT 2.1 User’s Guide. 2001. Fluent Inc.9. Nikander, K., J. Denyer, N. Smith, et al. 2001. Breath-

ing patterns and aerosol delivery: impact of regularhuman patterns, and sine and square waveforms onrate of delivery. J. Aerosol Med. 14:327:333.

10. FLUENT 6.1 User’s Guide. 2003. Fluent Inc.11. Katoshevski, D., and Y. Tambour. 1993. A theoretical

study of polydisperse liquid-sprays in a shear-layerflow. Phys. Fluids A 5:3085–3098.

12. Broday, D.M., and P.G. Georgopoulos. 2001. Growthand deposition of respirable hygroscopic particulatematter in the human lungs. Aerosol Sci. Technol.34:144–159.

13. Broday, D.M., and R. Robinson. 2003. Application ofcloud dynamics to dosimetry of cigarette smoke par-ticles in the lungs. Aerosol Sci. Technol. 37:510–527.

Received on January 16, 2006in final form March 22, 2006

Reviewed by:Chia-Ping Yu, Ph.D.

Asgharian Bahman, M.D.

Address reprint requests to:David Katoshevski, Ph.D.

Biotech. and Enviro. Eng., Building # 39Ben-Gurion University

P.O. Box 653Beer-Sheva, Israel

E-mail: davidk@bgumail.bgu.ac

SHAKKED ET AL.542