Content uploaded by Liran Oren
Author content
All content in this area was uploaded by Liran Oren on Jan 03, 2024
Content may be subject to copyright.
Elias Sundstr€
om
1
Department of Otolaryngology-Head and
Neck Surgery,
University of Cincinnati,
231 Albert Sabin Way,
Cincinnati, OH 45267
e-mail: sundstes@uc.edu
Rehab Talat
Department of Otolaryngology-Head and
Neck Surgery,
University of Cincinnati,
231 Albert Sabin Way,
Cincinnati, OH 45267
Ahmad R. Sedaghat
Department of Otolaryngology-Head and
Neck Surgery,
University of Cincinnati,
231 Albert Sabin Way,
Cincinnati, OH 45267
Sid Khosla
Department of Otolaryngology-Head and
Neck Surgery,
University of Cincinnati,
231 Albert Sabin Way,
Cincinnati, OH 45267
Liran Oren
Department of Otolaryngology-Head and
Neck Surgery,
University of Cincinnati,
231 Albert Sabin Way,
Cincinnati, OH 45267
Computational Modeling of
Nasal Drug Delivery Using
Different Intranasal
Corticosteroid Sprays for the
Treatment of Eustachian Tube
Dysfunction
Eustachian tube dysfunction (ETD) is a common otolaryngologic condition associated
with decreased quality of life. The first-line treatment of ETD is intranasal corticosteroid
sprays (INCS). Computational fluid dynamics (CFD) was used to study particle deposi-
tion on the Eustachian tube (ET) using two commercial INCS (Flonase and Sensimist).
Simulations also considered the effects of nostril side, insertion depth, insertion angle,
cone spray angle, inhaling rates, wall impingement treatment, and fluid film. Flonase and
Sensimist produced different particle size distributions and sizes. Sensimist droplets are
smaller, less sensitive to asymmetry in nostrils anatomy and variation in insertion angle,
and therefore can reach the posterior nasopharynx more readily. Flonase produces
larger particles with greater inertia. Its particles deposition is more sensitive to intrasub-
ject variation in nasal anatomy and insertion angles. The particle deposition on the ET
was sensitive to the wall impingement model. The deposition on the ET was insignificant
with adherence only <0.15%but increased up to 1–4%when including additional out-
comes rebound and splash effects when droplets impact with the wall. The dose redistrib-
ution with the fluid film is significant but plays a secondary effect on the ET deposition.
Flonase aligned parallel with the hard palate produced 4%deposition efficiency on the
ET, but this decreased <0.14%at the higher insertion angle. INCS with larger droplet
sizes with a small insertion angle may be more effective at targeting droplet deposition
on the ET opening. [DOI: 10.1115/1.4053907]
Keywords: computational fluid dynamics, nasal droplet deposition, spray injection
1 Introduction
Eustachian tube dysfunction (ETD) and its associated morbid-
ities of otitis media with effusion, tympanic membrane retraction,
and even cholesteatoma affect an estimated 11 million people per
year in the United States [1]. Intranasal corticosteroid sprays
(INCS) are considered the first-line treatment for ETD. This treat-
ment aims to deposit droplet particles on the Eustachian tube’s
(ET) tori tubarii aspect. A previous study has shown that nasal
anatomy (such as turbinates and septal deviations) can affect drop-
let deposition on the ET aspect [2], but the mechanisms of particle
deposition were not considered.
In the past decade, computational fluid dynamics (CFD) has
evolved into an essential research tool providing high spatio-
temporal resolution of nasal airflow and particle deposition in
patient-specific airway geometries. It has enabled a physics-based
understanding of droplet deposition. CFD has been used exten-
sively in both idealized and patient-specific geometries to study
drug delivery performance [3,4]. In the upper airway, numerous
studies have investigated the deposition of intranasal sprays in the
nasal cavity and paranasal sinuses [3–7]. However, these past
CFD studies did not specifically investigate particle deposition on
the ET and very few consider variations in the physical properties
of different INCS and their respective pump sprays.
Due to the many parameters that may influence the particle dep-
osition from INCS in the nasal cavity, it is crucial to adopt an
accurate computational approach with a fast turnaround time. Sev-
eral studies have employed the Reynolds-averaged Navier–Stokes
(RANS) equations with turbulence closure models to govern the
transport of the mean flow quantities [8,9]. Li et al. [10] per-
formed systematic numerical assessments comparing different
RANS turbulence closure formulations. They found that two-
equation models such as K–xSST can give comparable results of
mean flow quantities to more computational expensive large eddy
simulation (LES), an approach that is challenging for broad para-
metric deposition studies [11].
In this study, we use CFD modeling to investigate the deposi-
tion of INCS droplets on the ET opening. We simulate the spray
from two commercially available INCS, namely, Flonase and Sen-
simist, to predict particle deposition on the ET in patient-specific
models. The INCS nasal position (depth and angle) was also con-
sidered. Recent work on nasal spray deposition suggests that
agglomerated particles can develop into a liquid fluid film, which
in turn can undergo significant translocation to the posterior naso-
pharynx due to shear force interaction with the airflow and the
gravitational force [12]. Similarly, the wall impingement model in
this study assumes that particles will stick or adhere upon impact
with the mucosal layer in the nasal airway and then subsequently
develop a fluid film. One criticism with assuming adherence is
that the particle incident Weber number, defined as
1
Corresponding author.
Contributed by the Materials Division of ASME for publication in the JOURNAL OF
ENGINEERING AND SCIENCE IN MEDICAL DIAGNOSTICS AND THERAPY. Manuscript received
November 22, 2021; final manuscript received February 16, 2022; published online
March 11, 2022. Assoc. Editor: Mihai Mihaescu.
Journal of Engineering and Science
in Medical Diagnostics and Therapy
AUGUST 2022, Vol. 5 / 031103-1
Copyright V
C2022 by ASME
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
WeI¼qPv2
r;nDP=r, must be sufficiently low (less or equal than
two). Here, qP—is the particle density, vr;n—is the normal compo-
nent of the particle velocity relative to the wall, DP—is the parti-
cle diameter, and r—is the surface tension. If the effect of
particles with higher incident Weber numbers is not accounted
for, it could lead to a significant underestimation of particles
reaching the posterior nasopharynx. Therefore, this study also
aims to assess the particle deposition efficiency on the ET with
the Bai–Gosman wall impingement model that predicts different
behavior of liquid particles that impact the mucosal layer of the
nasal airway. Specifically, this model attempts to predict how and
when particles break up or adhere to the wall [13].
2 Methods
2.1 Model Geometries. Maxillofacial computed tomography
(CT) scans were acquired for two subjects. Each CT slice had
0.63-mm thickness with a spatial resolution of 0.43 mm/pixel in
each image (Fig. 1). Subjects 1 and 2 were adult female and male,
respectively. Both subjects had normal ET function. Demographic
information, e.g., age, weight, diagnosis, had been anonymized
and the study was deemed exempt by the internal review board at
the University of Cincinnati.
The control volume of the nasal airway geometries was
delineated from the CT scans using the medical imaging software
three-dimensional Slicer [14]. The inlet boundary of the nasal air-
way was specified externally of the nasal valve (Fig. 2). The outlet
was defined at the level of the hypopharynx. All other surfaces of
the control volume were specified as no-slip. The length, height,
and width of the nasal cavity for both subjects are listed in Table
1. The overall length scales (i.e., x
nasal
,y
nasal
,z
nasal
) show that the
subjects have similar nasal airway dimensions.
2.2 Model Parameters. Different constant inhalation flow
rates (Q) between 10 and 90 l/min were considered. These values
cover the range of normal adult inspiratory flow [15]. The
flow was assumed incompressible with a constant air density of
1.2 kg/m
3
and dynamic viscosity of 1.8 10
5
Pas. A turbulent
length scale of 7% to the equivalent diameter and a turbulent
intensity of 1% were specified at the inlet.
The CFD model considered the RANS for the incompressible,
steady, and viscous flow field [8,16,17]. The K–xSST model with
all yþwall treatment was used for turbulence closure [9,18]. A
finite volume upwind scheme of second order was used for
discretizing convective and diffusion terms in the governing
equations. The segregated flow solver in the CFD software
SIMCENTER STAR-CCMþwas used to control the solution coupling of
pressure and velocity fields according to the semi-implicit method
for pressure linked equations algorithm [8].
The liquid droplets from the INCS were treated as a dispersed
phase using the Lagrangian multiphase method [19]. Droplets’
transmission and dispersion in the airways were assumed to be
affected by turbulence, evaporation, atomization, and droplet break-
up. Their drag and shear lift forces were simulated using the
Schiller–Naumann correlation and the Sommerfeld correction,
respectively. The Ranz–Marshall correlation was used to calculate
the Sherwood number for regulation of the droplet evaporation [20].
A primary atomization distribution was assumed at the spray tip
with both INCS. Secondary droplet break-up was governed using
the Taylor-analogy distortion model with a critical Weber number
specified to 12 [21]. The simulations assumed that the droplets
remained spherical, had no rotational influence, and had negligible
particle-to-particle collision effects. The collision of a droplet with
the nasal airway wall was simulated using different wall impinge-
ment treatments. These included adherence only and fluid film treat-
ments. The difference is that liquid droplets merge with the mucosal
layer in the nasal airway with the former, while fluid film forms by
the droplets deposited on the wall with the latter. The fluid film can
subsequently translocate in the airway [12]. Simulations also con-
sidered the Bai–Gosman wall impingement treatment [13].
Fig. 1 CT scans of subject 1 (female, top row) and subject 2 (male, bottom row). Axial, sagittal, and
coronal planes are shown in the left to right columns. Images from subject 1 are annotated with the
different length scales.
031103-2 / Vol. 5, AUGUST 2022 Transactions of the ASME
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
The particle diameter distribution was specified to mimic the
characteristics of Flonase and Sensimist. These INCS were inves-
tigated by Hosseini et al. [22], and key characteristics used in this
study are given in Table 2. Their main difference is a smaller cone
angle of 20 deg with the Flonase, which also inject larger sized
particles (dp;mean) at a larger standard deviation (rp). The droplets
were simulated as liquid with a density of 1000 kg/m
3
. Both INCS
consider a single dose of 0.1 ml during an actuation time of
100 ms at a temperature of 25 C, corresponding to droplet counts
of 1 10
5
and 1 10
6
with the Flonase and Sensimist, respec-
tively. Droplets ejected into the nostrils were in an atomized state,
but an explicit primary atomization model was not used anywhere
else. The governing equations for the dispersed Lagrangian phase
were solved with 2
15
parcels, where each parcel represents a local-
ized group of droplets having the same properties. The cases with
fluid film treatment were simulated unsteady with the time-step
size 3 10
5
s, resulting in a Courant number around unity.
The INCS nasal position considered three cases for the insertion
angle (0 deg, 23 deg, and 43 deg), where a¼0 deg was defined as
injecting parallel to the hard palate. The nasal position also con-
sidered three insertion depths (5 mm, 0, and þ5 mm) where
0 mm was set in the centroid of the nasal vestibule (c.f. Fig. 2).
2.3 Model Verification and Validation. The control volume
of the nasal airway model was discretized with polyhedral cells
and prismatic cells next to no-slip wall boundaries. The pressure
drop between the inlet and the outlet was computed using three
Table 1 Nasal geometrical length scales, c.f. Fig. 1for the loca-
tion where the length scales are measured
(mm) Subject 1 (F) Subject 2 (M)
x
nasal
88 88
y
nasal
43 40
z
nasal
35 33
z
turb
29 32
z
w
1.7 1.4
Table 2 Spray cone injector conditions
Flonase Sensimist
Outer cone angle, h(deg) 20 35
Flow rate, Q
p
(ml/s) 1.1 1.1
Velocity magnitude (m/s) 14.5 14.4
Droplet distribution (lm) Normal, dp;mean ¼126;rp¼45 Log-normal, dp;mean ¼57;rp¼33
Fig. 2 Reconstruction of the model’s control volume geometry based on the CT scans from subject
1. (a) Stack of sagittal planes where the red frame highlights the model’s geometry region. (b) three-
dimensional control volume with inlet and outlet boundary conditions annotated with black arrows.
The spray cone injector is annotated with a dark gray triangle. The shaded gray area marks the surface
used for counting particles depositing on the ET. (c) Axial image indicating the location of the ET
opening. AA0is a plane located just upstream of the entrance to the ET and is used for assessing the
count of particles that can be deposited into the ET.
Fig. 3 Grid sensitivity analysis based on fine, medium, and coarse grids. (a) Variation of the pressure
drop as a function of the grid size. Richardson’s second-order extrapolation is shown with a dashed line.
(b) The numerical error for the infinite grid solution. (c) Pressure drop variation as a function of the flow
rate. The nasal airway replica measurement from Kelly et al. [23] is included for comparison.
Journal of Engineering and Science
in Medical Diagnostics and Therapy
AUGUST 2022, Vol. 5 / 031103-3
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
different grids (coarse, medium, and fine) (Fig. 3(a)). The coarse
grid used an averaged cell edge length of 2 mm, the medium grid
used 1 mm, and the fine grid used 0.5 mm. Ten prism layers with
stretching of 1.5 and thickness of 0.5 mm were used with all grids
to yield a wall yþaround unity. The finer grid was found to have
the least amount of error. Moreover, the error was seen to reduce
two orders of magnitude for each grid refinement order
(Fig. 3(b)). Therefore, the fine grid was used for the subsequent
analysis.
Figure 3(c)shows the computed pressure drop as a function of
the inhalation flow rate. The measurement data from a human
nasal airway replica from Kelly et al. [23] are included for com-
parison. Both the numerical data and the measurement show a
nonlinear dependence on the inhalation flow rate.
3 Results
The result section begins with using the simpler adhere wall
impingement treatment to assess the effects of different insertion
angles (Figs. 5and 6), the inspiratory flow rates (Fig. 7), and the
insertion depth (Fig. 8), and spraying in either nostril (Fig. 9).
Next, the effect of using different options for wall treatment is
shown (Figs. 10 and 11).
3.1 Effect of Droplet Evaporation. Flonase spray has an
overall longer penetration length than the Sensimist spray. Pene-
tration length is defined as the length from the initial spray to
when the particle evaporates. Liquid particles steadily lose mass
as they are converted into a vapor phase. Therefore, evaporation,
and the rate of mass being converted to vapor, is a mechanism
that limits how far the particles will travel. When the Flonase is
sprayed in the open atmosphere, the simulated penetration length
is in the order of 0.5 m and with the Sensimist, it is close to
0.25 m (Fig. 4). These length scales are in good agreement with
experimental data by Hosseini et al. [22].
The length scale (x
nasal
) of the nasal airway is about 88 mm for
subjects 1 and 2 (c.f. Table 1), which is 1/3rd of the length scale
of evaporation. Therefore, the effect of evaporation in the nasal
cavity is considered to be of second-order magnitude.
Figure 4(b)shows that the Weber number distribution of the
particle scatter is below its critical value for both INCS. Thus, the
Fig. 5 Particle tracks for three different insertion angles. Flonase to the left and Sensimist to the
right. Subject 1 is on the top row and subject 2 is on the bottom row. The inspiratory flow rate is
30 l/min.
Fig. 4 Penetration length for the Flonase (top) and the Sensimist (bottom). These scatter plots show how far droplet
particles would travel from the spray pumps exit outside the airway model under ambient atmospheric conditions.
The particles are scaled by the particle diameter and colored by the (a) particle diameter and (b) Weber number.
031103-4 / Vol. 5, AUGUST 2022 Transactions of the ASME
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
Fig. 6 Particle count (frequency) versus the particle diameter. Distribution is shown for the INCS exit (white
bars) and at section AA0for different insertion angles. Subject 1 is on the top row, and subject 2 is on the bottom
row. The Flonase is on the left column and Sensimist to the right.
Fig. 7 Particle count (frequency) versus the particle diameter at section AA0for different inspiratory flow rates.
(a) Flonase and (b) Sensimist.
Fig. 8 Particle count (frequency) versus the particle diameter at section AA0for different insertion depths of
the spray injector: (a) subject 1 and (b) subject 2. All cases are with Sensimist.
Journal of Engineering and Science
in Medical Diagnostics and Therapy
AUGUST 2022, Vol. 5 / 031103-5
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
fine spray of particles is considered to be in its atomized state and
no further break-up of droplets is expected.
3.2 Effect Intranasal Corticosteroid Sprays Insertion
Angle. The insertion angle of the INCS has a significant effect on
how droplets are convected in the nasal airway. The difference can
be appreciated qualitatively by tracking the particles at three differ-
ent insertion angles. The baseline case assumes the droplets are
injected at a¼0deg, from the x
nasal
(c.f. Fig. 2(b)for the definition
of the injection angle a). The other two cases simulate the particles
injected at 23deg, and 46 deg from the baseline. These angles are
within the possible administration angle range 0–90deg suggested
by Si et al. [12]. When the spray is injected approximately parallel
to the hard palate (a¼0 deg), more particle tracks reach the poste-
rior nasal cavity. In both models, the higher the insertion angles,
the particle tracks deposit more anteriorly (Fig. 5).
The histogram in Fig. 6quantifies the particle count at plane
AA0for all considered cases. Plane AA0is located just upstream
of the ET (c.f. Figs. 2(a)and 2(c)) and can account for all particles
passing by the ET entrance. The columns show the particle count
at AA0and compare it with the count at the INCS exit. It shows
that Flonase produces larger sized particles with a normal distribu-
tion (Figs. 6(a)and 6(c)).
Independent of the insertion angle, most droplet particles from
Flonase are deposited in the nasal cavity before reaching AA0.
This can also be appreciated in the particle tracking shown in
Fig. 5. The simulation shows that when the spray is injected more
parallel to the hard palate, up to 30% of the injected particles in
subject 1 (Fig. 6(a)) may penetrate posteriorly, whereas with a 23-
deg insertion angle, almost all injected particles deposit before
reaching the posterior nasal cavity. The deposition efficiency can
be defined as the quotient between the number of particles hitting
a specific surface (N
surf
) to the total number of particles injected
by the INCS (N
tot
)as
gp¼Nsurf =Ntot (1)
It shows that deposition efficiency was higher for subject 1 than
subject 2. In subject 2, only up to 10% of the injected particles
may penetrate posteriorly (Fig. 6(c)). Thus, the particle deposition
efficiency is seen to be sensitive to the patient-specific geometry.
The Sensimist spray produces smaller particle sizes with a log-
normal distribution at its exit and smaller particle size variability
than Flonase (Figs. 6(b)and 6(d)). As such, smaller sized particles
are also deposited posteriorly. The particles observed at AA0have
a mean particle diameter dp;mean ¼18 lm. The particle deposition
decreases when the INCS insertion angle is increased. However,
Sensimist is seen to be less sensitive to an increase of the insertion
angle than Flonase. The quantitative difference in deposition
between the two spray pumps is signified by particle inertia. Small
inertia indicates that particles can more easily adjust their spatial
trajectory to follow the streamlines of the inspiratory airflow.
They can, therefore, be carried by the inspiratory airflow around
relatively complex and highly curved geometry and reach the pos-
terior regions of the nasal cavity and the nasopharynx. In contrast,
the larger Flonase particles with higher inertia are lesser influ-
enced by the inspiratory airflow and deposit more anteriorly,
unless the Flonase insertion angle is parallel with the hard palate
(a¼0 deg).
3.3 Variation Due to Inspiratory Flow Rate. Figure 7
shows the sensitivity of the particle count at section AA0for varia-
tion of the flow rate. It shows that Flonase slightly depends on the
inspiratory flow rate, whereas a more significant variation is
observed with Sensimist. This variation affects particle sizes with
a mean diameter of around 18 lm (c.f. Figs. 6(c)and 6(d)). For
lower inspiratory flow rates, the particle deposition posteriorly
increases with Sensimist. Plots are shown only for Subject 1, but
similar results were observed for subject 2.
3.4 Effect of Intranasal Corticosteroid Sprays Insertion
Depth. Figure 8shows the variation in particle count at section
AA0based on the insertion depth. The insertion angle was kept at
23 deg and the inspiratory flow rate was held constant at 30 l/min.
It shows that deposition in the posterior nasal cavity increases
with the Sensimist when the spray tip is further inserted into the
nose. This observation is consistent for both subjects.
Fig. 9 Difference in particle count (frequency) as a function of particle diameter at section AA’ when injecting
in each nostril: (a) subject 1 and (b) subject 2. All cases are with Sensimist.
Fig. 10 Fluid film thickness evolution in the nasal airway for subject 1. A dose of 0.1 ml is
sprayed using Flonase with injection angle a523 deg and inspiratory flow rate 30 l/min.
031103-6 / Vol. 5, AUGUST 2022 Transactions of the ASME
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
3.5 Intrasubject Variation Due to Nasal Anatomy. All pre-
vious results were based on the INCS injected inside the left nos-
tril of either subject. The effect of spraying in the right nostril was
subject-specific (Fig. 9). The data shown are based on the Sensi-
mist injected at 23-deg insertion angle with 30 l/min of inspiratory
flow. In subject 1, nearly 50% fewer particles of sizes below
100 lm were observed in section AA0when spraying inside the
right nostril. On the other hand, there was little difference in
patient 2 when spraying in either nostril. The difference between
subjects probably stems from the asymmetry in the left and right
nasal passages, which is more prominent in subject 1.
3.6 Effect of Fluid Film Translocation on the Nasal Air-
way Wall. The effect of fluid film formation from droplets adher-
ing and agglomerating is shown in Fig. 10. This case considers a
dose of 0.1 ml sprayed with Flonase at 23-deg insertion angle and
30 l/min inspiratory flow. The figure shows several instantaneous
snapshots with different levels of development and redistribution
of the fluid film. In the first snapshot (t¼2 ms), the spray has
penetrated the nasal passage and a small thin (thickness 5 lm)
patch of fluid film is observed. A significant fluid film develop-
ment occurs until t¼10 ms, where the initial patch has grown and
a second isolated patch is developing. Toward snapshot t¼90 ms,
just before the end of the dose administration, the fluid film covers
a large area of the middle section of the nasal airway, and the film
thickness is now around 200 lm. After the dose administration
(t¼100 ms), the fluid film mass remains constant and redistributes
under the gravitational force and the interaction of shear forces
with nasal airflow. The result of this interaction is shown at
t¼150 ms. The fluid film has redistributed posteriorly and inferi-
orly toward the soft palate. The fluid film redistribution from
t¼90 ms until t¼400 ms is in the order of a few centimeters per
second. This velocity is two orders of magnitude slower than the
velocity scale of the liquid droplets (c.f. Table 2). Another inter-
esting observation is that the fluid film redistribution does not pass
across the entrance region to the ET, but instead flows more inferi-
orly, which leads to poor deposition efficiency on the ET.
3.7 Effect of Advanced Wall Impingement Model. In the
result shown so far, the wall impingement treatment assumed that
the liquid droplets adhere upon collision with the nasal airway
wall. If droplet behavior such as rebound and splash is also con-
sidered (using the Bai–Gosman wall impingement model), particle
distribution changes because of the higher incident Weber number
(Fig. 11). With Flonase aligned parallel with the hard palate,
almost 75% enters the posterior nasal airway (Fig. 11(a)) com-
pared with nearly 55% with Sensimist (Fig. 11(b)). These values
are higher than the 30% and 20% for the Flonase and Sensismist,
respectively, previously predicted. The Flonase shows a reduced
particle count at the higher insertion angle, resulting in poor
efficiency. Sensimist shows a relatively higher particle count indi-
cating that the insertion angle affects the droplet count at AA0
less.
The simulations also show that only a fraction of the observed
particles in the posterior nasal cavity (i.e., AA0plane) get specifi-
cally deposited on the ET. This observation is made by calculating
the deposition efficiency (Eq. (1)) on the ET surface. In both sub-
jects, deposition efficiency on the ET was in the order of 1% for
both sprays (Fig. 12). These observations are made with the INCS
injected in the left nostril at insertion depth Dx¼0 mm, and
inspiratory flow 30 l/min (Fig. 12(a)). The adherence-only wall
impingement model is even lower (<0.15%). Flonase is sensitive
to the insertion angle, with a few percent deposition efficiencies at
the smaller insertion angles and almost no efficiency at the larger
46 deg angle. Flonase also shows variability between the two sub-
jects when considering the insertion angle and the insertion depth.
Sensimist shows a smaller sensitivity to the insertion angle and
between the subjects. Although it has a lower efficiency for inser-
tion angle 0 deg than the Flonase, its efficiency is better at higher
insertion angles. Figure 12(b)shows that the particle deposition
efficiency on the ET consistently increases with the inspiratory
flow for all cases. Figure 12(c)shows the deposition efficiency
with Sensimist increases in both subjects. However, the effect of
insertion depth with Flonase was subject-specific. Specifically, the
efficiency seemed to vary more and have an optimal insertion
depth with subject 1, but it did not change much with subject 2.
4 Discussion
This study showed that particle deposition in the posterior nasal
cavity varies with INCS. Specifically, particle deposition at the
entrance of the Eustachian tube is more favorable when Flonase is
aligned parallel with the hard palate. On the other hand, Sensimist
shows less variability in terms of the insertion angle and between
the two patient-specific geometries. Varying the inspiratory flow
rate, insertion angle, cone angle, or insertion depth does not
change the deposition efficiency much. The wall impingement
model has a significant effect on the deposition. With adherence,
which is commonly used in previous CFD studies on nasal drug
delivery, the particle deposition efficiency on the Eustachian tube
was below 0.15% in all considered cases. With a more advanced
impingement, which includes rebound and splash of the droplets
from the wall, the efficiency increases to a few percent. Most par-
ticles injected into the nasal cavity will pass by the entrance to the
Eustachian tube without being deposited on it. These observations
suggest that other spray pumps should be considered when the
therapy targets drug delivery to the ET using Flonase or
Sensimist.
The deposition mechanism for droplets on the ET remains
poorly characterized despite the frequent use of INCS to treat
ETD. In this study, we sought to determine the deposition profiles
Fig. 11 Particle count (frequency) versus the particle diameter at AA0for different insertion angles using the
Bai–Gosman wall impingement model for subject 1
Journal of Engineering and Science
in Medical Diagnostics and Therapy
AUGUST 2022, Vol. 5 / 031103-7
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
for patient-specific geometry using two spray pumps that are com-
mercially available for patients. We found that the wall impinge-
ment model significantly influences the deposition efficiency of
the ET. With a simple adherence impingement model, the deposi-
tion efficiency is very low on the ET but increases to a few per-
cent when the Bai-Gosman impingement model was used. We
found that Sensimist, with its small particle size and low inertia,
shows less variability to the insertion angle that may give an over-
all higher probability of reaching the posterior nasal cavity and
depositing in the nasopharynx than Flonase. However, it was
found that a significant number of particles can reach the posterior
nasal cavity if the Flonase spray is injected roughly parallel to the
hard palate.
Previous studies have often relied on CFD modeling techniques
to characterize mechanisms of nasal drug delivery. A major limi-
tation occurs when studies simplify nasal flow parameters to run
the CFD simulation. As such, their findings may not accurately
represent the mechanisms that exist in vivo, as discussed in the
review by Zubair [24]. However, more recent studies have
included the patient-specific details of the nasal anatomy, resting
breathing rates, and both laminar and turbulent flow. Results can
also vary when studies assess different turbulence models with
RANS when studying particle deposition in the nasal cavity at
resting flow rates [3].
In contrast to previous studies, our study focused on particles
depositing on the Eustachian tube and considered the effects of
different spray cone angles, insertion angles, inhaling rates, inser-
tion depths, and patient-specific nasal anatomy. Our study also
includes the effect of fluid film translocation, which was sug-
gested to affect dose redistribution to the olfactory [12]. It pre-
dicted lower deposition efficiency if fluid film develops because
of its pathway in the nasal cavity. However, considering the effect
of rebound and splash of droplets from the nasal walls predicted a
more significant deposition efficiency on the ET than fluid film
alone.
The particle Weber number governs the primary atomization
and secondary break-up. The Weber number refers to the ratio
between inertial forces and surface tension and is defined as the
air density times the square of the particle velocity times the parti-
cle diameter divided by the surface tension. Primary atomization
is the mechanism of forcing a liquid through a narrow orifice at a
high pressure that produces an atomized spray of small droplets
[25]. Secondary break-up refers to the process of liquid droplets
breaking up under the influence of nonuniform surfaces forces
that occur in the interaction with the continuous gas phase
[26,27]. According to the study by Huh et al. [28], a particle will
not undergo the primary atomization stage if the Weber number is
less than 12. Moreover, a secondary break-up is also bypassed for
subcritical Weber numbers [27], where a Weber number of 12 is
critical for a vibrational break-up to manifest. The simulated result
showed that the Weber number of the particle cloud with both
INCS was less than 12. Hence, it was concluded that the particles
are injected in an atomized state with no need for an explicit pri-
mary atomization model.
4.1 Study Limitations. Using a small cohort is a limitation
that was needed to reduce the number of unknown geometric fac-
tors that may influence the result. Examining intra- and intersub-
ject variability in further detail will require are larger cohort. We
also used the RANS equations in our CFD modeling, and a future
study could benefit from comparing with data generated using
LES [11,29,30].
Realistic simulations of spray would describe how a continuous
Eulerian liquid phase inside the INCS is pushed through a narrow
orifice and break up into a scatter of discrete Lagrangian droplets.
However, due to the large separation in physical scales, droplets
of a few micrometers compared to centimeters of the liquid col-
umn in the INCS, make a full Eulerian description impractical.
Instead, this study is limited by the assumption that particles
ejected from the INCS are already in an atomized state with a
specified particle size distribution. This assumed discrete spherical
particles with negligible rotational influence, and negligible
particle–particle interaction. The benefit was that no explicit pri-
mary atomization was needed to account for the complex physics
inside the nozzle where continuous liquid break-up into droplets.
In general, near the nozzle of the INCS there is a small volume
space with significant interaction of continuous phase and dis-
persed phase. This area with densely packed particles where the
effect of interparticle interaction is important could be treated
with additional contact forces, which enter the Lagrangian
momentum equations.
5 Conclusions
In conclusion, a computational model for nasal drug delivery
using different intranasal corticosteroid sprays for the treatment of
Eustachian tube dysfunction was developed. The numerical simu-
lations assessed the particle deposition efficiency on the ET con-
sidering the effects of intra- and intersubject variations, spraying
in either nostril side, insertion depth, insertion angle, inspiration
rates, wall impingement treatment, and fluid film. The main find-
ings are:
(I) Particle deposition efficiency on the ET is less sensitive
with Sensimist to intra- and inter-subject anatomy and
variations in insertion angle.
Fig. 12 Particle deposition efficiency at the ET as a function of the (a) spray insertion angle. Data is shown for both
subjects and the Flonase and Sensimist spray pumps. (b) Variation to the flow rate sprayed with insertion
angle 23 deg. (c) Variation to the insertion depth for Sensimist with insertion angle 0 deg and inspiratory flow
30 l/min.
031103-8 / Vol. 5, AUGUST 2022 Transactions of the ASME
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
(II) Particle deposition efficiency on the ET is more sensitive
with Flonase to intra-subject variation in the nasal anat-
omy and insertion angles.
(III) Both INCS show increased particle deposition efficiency
on the ET when including additional outcomes rebound
and splash effects in the wall impingement model,
whereas fluid film shows an insignificant effect.
(IV) INCS with larger droplet sizes and an insertional angle
roughly parallel with the hard palate may yield a more
effective particle deposition on the ET.
Funding Data
NIH (Grant No. K25DC014755; Funder ID: 10.13039/
100000002).
Conflict of Interest
There are no conflicts of interest.
References
[1] Shan, A., Ward, B. K., Goman, A. M., Betz, J. F., Reed, N. S., Poe, D. S., and
Nieman, C. L., 2019, “Prevalence of Eustachian Tube Dysfunction in Adults in
the United States,” JAMA Otolaryngol. Head Neck Surg.,145(10), p. 974.
[2] Aggarwal, R., Cardozo, A., and Homer, J. J., 2004, “The Assessment of Topical
Nasal Drug Distribution,” Clin. Otolaryngol. Allied Sci.,29(3), pp. 201–205.
[3] Liu, Y., Matida, E. A., Gu, J., and Johnson, M. R., 2007, “Nu merical Simulation
of Aerosol Deposition in a 3-D Human Nasal Cavity Using RANS, RANS/
EIM, and LES,” J. Aerosol Sci.,38(7), pp. 683–700.
[4] Xi, J., Kim, J., and Si, X. A., 2016, “Effects of Nostril Orientation on Airflow
Dynamics, Heat Exchange, and Particle Depositions in Human Noses,” Eur. J.
Mech. B/Fluids,55, pp. 215–228.
[5] Kleven, M., Melaaen, M. C., Reimers, M., Røtnes, J. S., Aurdal, L., and Djupes-
land, P. G., 2005, “Using Computational Fluid Dynamics (CFD) to Improve the
Bi-Directional Nasal Drug Delivery Concept,” Food Bioprod. Process.,83(2),
pp. 107–117.
[6] Kimbell, J. S., Segal, R. A., Asgharian, B., Wong, B. A., Schroeter, J. D.,
Southall, J. P., Dickens, C. J., Brace, G., and Miller, F. J., 2007,
“Characterization of Deposition From Nasal Spray Devices Using a Computa-
tional Fluid Dynamics Model of the Human Nasal Passages,” J. Aero sol Med.,
20(1), pp. 59–74.
[7] Matida, E. A., Finlay, W. H., Lange, C. F., and Grgic, B., 2004, “Improved
Numerical Simulation of Aerosol Deposition in an Idealized Mouth–Throat,” J.
Aerosol Sci.,35(1), pp. 1–19.
[8] Ferziger, J. H., and Peric´, M., 1996, Computational Methods for Fluid Dynam-
ics, Springer, Berlin Heidelberg, Berlin, Heidelberg, Germany.
[9] Pope, S. B., 2001, Turbulent Flows, Cambridge University Press, New York.
[10] Li, C., Jiang, J., Dong, H., and Zhao, K., 2017, “Computational Modeling and
Validation of Human Nasal Airflow Under Various Breathing Conditions,” J.
Biomech.,64, pp. 59–68.
[11] Calmet, H., Inthavong, K., Eguzkitza, B., Lehmkuhl, O., Houzeaux, G.,
and V
azquez, M., 2019, “Nasal Sprayed Particle Deposition in a Human
Nasal Cavity Under Different Inhalation Conditions,” PLoS One,14(9),
p. e0221330.
[12] Si, X. A., Sami, M., and Xi, J., 2021, “Liquid Film Translocatio n Significantly
Enhances Nasal Spray Delivery to Olfactory Region: A Numerical Simulation
Study,” Pharm.,13(6), p. 903.
[13] Bai, C. X., Rusche, H., and Gosman, A. D., 2002, “Modeling of Gasoline Spray
Impingement,” Atom. Sprays,12(1–3), pp. 1–28.
[14] Fedorov, A., Beichel, R., Kalpathy-Cramer, J., Finet, J., Fillion-Robin, J. C.,
Pujol, S., Bauer, C., Jennings, D., Fennessy, F., Sonka, M., Buatti, J., Aylward,
S., Miller, J. V., Pieper, S., and Kikinis, R., 2012, “3D Slicer as an Image Com-
puting Platform for the Quantitative Imaging Network,” Magn. Resonan. Imag.,
30(9), pp. 1323–41.
[15] Mead, J., and Whittenberger, J. L., 1953, “Physical Properties of Human
Lungs Measured During Spontaneous Respiration,” J. Appl. Physiol.,5(12),
pp. 779–796.
[16] Sundstr€
om, E., and Oren, L., 2019, “Pharyngeal Flow Simulations During Sibi-
lant Sound in a Patient-Specific Model With Velopharyngeal Insufficiency,” J.
Acoust. Soc. Am.,145(5), pp. 3137–3145.
[17] Sundstr€
om, E., Boyce, S., and Oren, L., 2020, “Effects of Velopharyngeal
Openings on Flow Characteristics of Nasal Emission,” Biomech. Model.
Mechanobiol.,19(5), pp. 1447–1413.
[18] Menter, F. R., 1994, “Two-Equation Eddy-V iscosity Turbulence Models for
Engineering Applications,” AIAA J.,32(8), pp. 1598–1605.
[19] Crowe, C. T., Schwarzkopf, J. D., Sommerfeld, M., and Tsuji, Y., 2011, Multi-
phase Flows With Droplets and Particles, CRC Press, Boca Raton, FL.
[20] Ranz, W., and Marshall, W., 1952, “Evaporation From Drops. Parts I & II,”
Chem. Eng. Progr., 48, pp. 141–146.
[21] O’Rourke, P. J., and Amsden, A. A., 1987, “The Tab Method for Numerical
Calculation of Spray Droplet Breakup,” SAE Paper No. 872089.
[22] Hosseini, S., Wei, X., Wilkins, J. V., Fergusson, C. P., Mohammadi, R., Vor-
ona, G., and Golshahi, L., 2019, “In Vitro Measurement of Regional Nasal
Drug Delivery With Flonase,
V
R
Flonase
V
R
Sensimist,
TM
and MAD Nasal
TM
in
Anatomically Correct Nasal Airway Replicas of Pediatric and Adult Human
Subjects,” J. Aerosol Med. Pulm. Drug Deliv.,32(6), pp. 374–385.
[23] Kelly, J. T., Asgharian, B., Kimbell, J. S., and Wong, B. A., 2004, “Particle
Deposition in Human Nasal Airway Replicas Manufactured by Different
Methods. Part I: Inertial Regime Particles,” Aerosol Sci. Technol.,38(11),
pp. 1063–1071.
[24] Zubair, M., 2012, “Review: A Critical Overview of Limitations of Cfd Model-
ing in Nasal Airflow,” J. Med. Biol. Eng.,32(2), p. 77.
[25] Senecal, P., Schmidt, D., Nouar, I., Rutland, C., Reitz, R., and Corradini, M.,
1999, “Modeling High-Speed Viscous Liquid Sheet Atomization,” Int. J. Multi-
phase Flow,25(6–7), pp. 1073–1097.
[26] Reitz, R. D., and Diwakar, R., 1986, “Effect of Drop Breakup on Fuel Sprays,”
SAE Paper No. 860469.
[27] Stiesch, G., 2003, Modeling Engine Spray and Combustion Processes, Springer,
Berlin/Heidelberg, Germany.
[28] Huh, K. Y., Lee, E., and Koo, J., 1998, “Diesel Spray Atomization Model
Considering Nozzle Exit Turbulence Conditions,” Atom. Sprays,8(4),
pp. 453–469.
[29] Sundstr€
om, E., and Oren, L., 2019, “Sound Production Mechanisms of Audible
Nasal Emission During the Sibilant/S,” J. Acoust. Soc. Am.,146(6),
pp. 4199–4210.
[30] Sundstr€
om, E., and Oren, L., 2021, “Change in Aeroacoustic Sound Mechanism
During Sibilant Sound With Different Velopharyngeal Opening Sizes,” Med.
Biol. Eng. Comput.,59(4), pp. 937–945.
Journal of Engineering and Science
in Medical Diagnostics and Therapy
AUGUST 2022, Vol. 5 / 031103-9
Downloaded from http://asmedigitalcollection.asme.org/medicaldiagnostics/article-pdf/5/3/031103/6863544/jesmdt_005_03_031103.pdf by University of Cincinnati user on 03 January 2024
