ARTICLE IN PRESS

0021-9290/$ - se

doi:10.1016/j.jb

�CorrespondUniversity Hos

Tel.: +32 47538

E-mail addr1These autho

Journal of Biomechanics 40 (2007) 3708–3714

www.elsevier.com/locate/jbiomech

www.JBiomech.com

Functional imaging using computational fluid dynamics to predicttreatment success of mandibular advancement devices in

sleep-disordered breathing

J.W. De Backera,b,�,1, O.M. Vandervekenb,d,1, W.G. Vosa,b, A. Devolderb, S.L. Verhulstf,J.A. Verbraeckenb, P.M. Parizele, M.J. Braemc, P.H. Van de Heyningd, W.A. De Backerb

aDepartment of Physics, University of Antwerp, Antwerp, BelgiumbDepartment of Respiratory Medicine, University Hospital Antwerp, University of Antwerp, Antwerp, Belgium

cDepartment of Dentistry, University Hospital Antwerp, University of Antwerp, Antwerp, BelgiumdDepartment of ENT, Head and Neck Surgery, University Hospital Antwerp, University of Antwerp, Antwerp, Belgium

eDepartment of Radiology, University Hospital Antwerp, University of Antwerp, Antwerp, BelgiumfDepartment of Pediatrics, University Hospital Antwerp, University of Antwerp, Antwerp, Belgium

Accepted 19 June 2007

Abstract

Mandibular advancement devices (MADs) have emerged as a popular alternative for the treatment of sleep-disordered breathing.

These devices bring the mandibula forward in order to increase upper airway (UA) volume and prevent total UA collapse during sleep.

However, the precise mechanism of action appears to be quite complex and is not yet completely understood; this might explain

interindividual variation in treatment success. We examined whether an UA model, that combines imaging techniques and

computational fluid dynamics (CFD), allows for a prediction of the treatment outcome with MADs. Ten patients that were treated with

a custom-made mandibular advancement device (MAD), underwent split-night polysomnography. The morning after the sleep study, a

low radiation dose CT scan was scheduled with and without the MAD. The CT examinations allowed for a comparison between the

change in UA volume and the anatomical characteristics through the conversion to three-dimensional computer models. Furthermore,

the change in UA resistance could be calculated through flow simulations with CFD. Boundary conditions for the model such as mass

flow rate and pressure distributions were obtained during the split-night polysomnography. Therefore, the flow modeling was based on a

patient specific geometry and patient specific boundary conditions. The results indicated that a decrease in UA resistance and an increase

in UA volume correlate with both a clinical and an objective improvement. The results of this pilot study suggest that the outcome of

MAD treatment can be predicted using the described UA model.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Imaging; Oral appliances; Sleep apnea; Snoring; Upper airway; Computational fluid dynamics

1. Introduction

The awareness of the potentially serious consequences ofsleep related breathing disorders (SBD) such as obstructivesleep apnea hypopnea syndrome (OSAHS) is increasing.

e front matter r 2007 Elsevier Ltd. All rights reserved.

iomech.2007.06.022

ing author. Department of Respiratory Medicine,

pital Antwerp, Wilrijkstraat 10, 2650 Edegem, Belgium.

8786; fax: +3238214447.

ess: Jan.DeBacker@ua.ac.be (J.W. De Backer).

rs made equal contributions to this work.

OSAHS is characterized by recurrent episodes of partial orcomplete collapse of the upper airway. This may lead to afragmentation of the sleep pattern, a decrease in oxygensaturation and a partial pressure rise of CO2 in the blood.The arousals and the nocturnal hypoxemia can causeexcessive daytime sleepiness, loss of concentration, hyper-tension and atherosclerosis. In some cases, the pronouncedpresence of OSAHS may lead to stroke and even heartfailure, resulting in an increased prevalence of cardiovas-cular morbidity and mortality (De Backer, 2006; AmericanAcademy of Sleep Medicine Task Force, 1999). Nowadays,

ARTICLE IN PRESSJ.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–3714 3709

the golden standard to assess the severity of sleep apnea isthe apnea–hypopnea index (AHI). This index presents thenumber of UA narrowings and closures per hour. An AHIabove 5 events h�1 in combination with complaints isregarded as an OSAHS.

Mandibular advancement devices (MADs), which areworn intra-orally at night to advance the lower jaw, haveemerged as a non-invasive treatment for SBD (Marklundet al., 2004; Lim et al., 2006; Vanderveken et al., 2004).There is evidence that MADs can significantly reduce thecollapsibility of the upper airway (UA) during sleep (Katoet al., 2000; Ng et al., 2003; Huang et al., 2005). Thewidening of the pharyngeal cross-sectional area (CSA) asinduced by MADs can occur at the level of velo-, oro- and/or hypopharynx (Ryan et al., 1999; Kato et al., 2000;Tsuiki et al., 2004, 2005). Both the reduction in pharyngealcollapsibility and the widening of CSA, caused by MADs,have been reported to be dose-dependent (Kato et al., 2000;Tsuiki et al., 2005). However, in recent studies thatcompare different degrees of mandibular advancement, amore pronounced mandibular advancement did not lead toa greater improvement compared to less advancement forpatients with mild to moderate OSAHS (Tegelberg et al.,2003; Walker-Engstrom et al., 2003).

MAD therapy may be a first line treatment in snorerswithout (primary) or with (non-apneic) excessive daytimesleepiness and in patients with mild to moderate OSAHS(Kushida et al., 2006; Ferguson et al., 2006; Hoekemaet al., 2004). Treatment with MADs may also beconsidered in subjects with OSAHS who do not tolerateor comply with continuous positive airway pressure(CPAP) or as a temporary alternative (Lim et al., 2006;Smith and Stradling, 2002). Although promising, successrate with MADs is limited, and a high interindividualvariability has been noted (Marklund, 2006a). Therefore, itwould be worthwhile to have a tool to distinguishresponders from non-responders. However, the data inthe literature on predictability of treatment outcome withMADs are not conclusive and more research is needed todevelop a adequate prediction method (Cistulli andGotsopoulos, 2004; Cistulli et al., 2004). Recently, wehave developed an UA model based on the principles ofcomputational fluid dynamics (CFD) to develop an

Fig. 1. Custom-made mandib

adequate prediction tool for interventions in the UA.(Vos et al., 2007). Computed tomography (CT) data aretransformed into computational models and analyzedthrough CFD. In this paper, results of a pilot study usingthis UA model for the assessment of treatment success withMAD in SBD are discussed.

2. Material and methods

2.1. Patient data and mandibular advancement device

For this study 10 adult patients with heavy snoring and an

apnea–hypopnea index (AHI) o40 h�1 on standard polysomnography

(PSG) were selected. All of these patients were treated with a custom-made

MAD. The custom-made MAD used in this study (Fig. 1) was fabricated

by a dental technician based on individually made plaster casts of the teeth

and a bite registration taken by our dentist; the MAD was constructed at

the dental laboratory of the Umea University, Sweden (Marklund et al.,

1998, 2004; Marklund, 2006b). The study was performed according to

institutional guidelines and informed consent was obtained for all patients.

2.2. Polysomnographic study

All patients underwent three sleep studies. The first diagnostic PSG

confirmed an AHI below 40 h�1. After a 4-month treatment period a

similar standard PSG was performed with the MAD to determine

treatment efficacy. Standard full-night PSG (Medilog SAC, Oxford

Instruments, Oxon, UK) was performed. Finally, patients were invited

for a split-night sleep study with and without the MAD. These split-night

studies were performed using the Alice III system (Healthdyne Technol-

ogies, Marietta, GA, USA) and included standard PSG. Airflow was

detected by a Fleisch No. 1 pneumotachograph. Continuous monitoring

of respiratory effort was performed using the esophageal sensor of a multi-

sensor UA catheter as previously described (Boudewyns et al., 2001;

Vanderveken et al., 2005). Sleep recordings were scored manually in a

standard fashion (American Academy of Sleep Medicine Task Force,

1999; Levy et al., 2006).

2.3. Computational domain and grid

The morning after the split-night PSG, two CT scans (Siemens

Sensation 16 with a slice thickness of 0.5mm) were performed with and

without MAD. The scanning area started at the nasopharynx and

extended down to the larynx. Based on these CT images the maxilla,

mandibula and UA could be reconstructed into three-dimensional (3D)

computer-aided design (CAD) models. We used Mimicss software

(Materialise, Leuven, Belgium) for the construction of 3D models based

on Hounsfield units (HU). In order to reconstruct 3D images based on

ular advancement device.

ARTICLE IN PRESSJ.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–37143710

HU, the appropriate voxels had to be grouped accordingly. Therefore,

masks were created that contain voxels with predefined HU. Since we were

interested in constructing the maxilla, mandibula and upper airway, three

masks were created. Once the masks were defined accurately with the

appropriate HU bounds, [�1024, �400] for air and [226, 3071] for bone

structure, the 3D models could be constructed using triangulation. Once

the UA volume was converted into a 3D image, it could be used to analyze

the flow behavior inside the airway using CFD for both cases. To this end

a computational grid was created using TGrid 4.0.16 (Ansys, Lebanon,

USA). Depending on the complexity of the UA model a typical grid

consisted of 600,000–800,000 tetrahedral cells. A grid dependency showed

near identical results for increasing grid sizes.

2.4. Physical models and boundary conditions

Flow simulations were performed using the commercial code Fluent

v6.3 (Ansys, Lebanon, USA) that solved the Reynolds averaged

Navier–Stokes equations. The flow was assumed to be steady, homo-

geneous, incompressible, adiabatic and Newtonian. The pressure-based

solver was used with a node-based Green–Gauss gradient treatment. The

latter achieves higher accuracy in unstructured tetrahedral grids compared

to the cell-based gradient evaluation. A second-order pressure discretiza-

tion scheme was used for the pressure calculation and a second-order

upwind scheme was used for the momentum and turbulence transport

equations. The pressure–velocity coupling was solved using the SIMPLE

scheme. Due to the narrowness of the UA region, turbulent features

are to be expected in the model even at a Reynolds number of around

6000. Therefore, a two-equation model was used to represent the

flow turbulence. For this type of problem the low Reynolds number

(LRN) k–o has been used, since this model can accurately predict pressure

drops and velocity profiles. In addition, the LRN k–o model can

obtain an accurate laminar solution when the turbulent viscosity

approaches zero (Wilcox, 1998). A velocity profile was defined at

the inlet and a pressure at the outlet of the model. The mass flow

rate used in the model was taken equal to the mass flow rate measured

by the pneumotach. The outlet pressure for the model was taken equal to

the in-vivo pressure measurement during PSG obtained by a pressure

catheter.

Fig. 2. Change in mandibular position and upper airway morphology for a su

Upper panel: without MAD; lower panel: with MAD.

2.5. Outcome parameters

The angulus mandibulae (AM) (Putz, 1994) was assessed for all

patients. To determine this angle two construction lines were created. One

line connected the midpoint of the arc formed by the AM with the center

of the condylus. The other line contained the same midpoint and acted as a

tangent line to the lower part of the mandibula. The angle between both

lines was considered the AM. Fig. 2 illustrates the construction lines in

both a successfully and unsuccessfully treated patient. Besides the AM,

also the change in UA volume was calculated, as enlarging the UA volume

is the main goal of an MAD. The change was determined by comparing

the UA volumes extracted from the CT scans. To ensure that the

comparison was accurate, the UA model limits were identical for both

cases. This was done by matching the CT slice numbers. The CFD

calculations yielded the total pressure drop over the model which could

then be used to determine the airway resistance R:

R ¼Dp

Fua, (1)

where Dp is the total pressure drop and Fua the mass flow rate in the upper

airway.

In this study, the clinical improvement is defined as a decrease in

snoring and/or a decrease in AHI below 50% compared to baseline.

Statistical analysis was performed using the Spearman correlation

coefficient (Statistica 7.0, Statsoft). A comparison between responders

and non-responders was made using a Mann–Whitney U-test with p-

values based on the exact distribution. For all analyses, po0.05 was

considered statistically significant.

3. Results

An overview of the patient data and the results of PSGand UA modeling are given in Table 1.The PSG showed for all 10 patients an average AHI of

17.2710.3. The MAD treatment was unsuccessful in 3 outof 10 patients (33.3%). The successfully treated patients

ccessfully (left panel) and an unsuccessfully (right panel) treated patients.

ARTICLE IN PRESS

Table 1

Overview of baseline characteristics and data from the upper airway model for all patients

Patient Age Sex BMI AHI without

MAD

AHI with

MAD

Angulus mandibulae

(deg)

Change in upper airway

volume (%)

Change in upper airway

resistance (%)

Treatment

success

1 48 M 27 24 1 120 28 �56 Yes

2 60 M 34 31 9 118 204 n/a Yes

3 57 M 25 14 2 110 55 �53 Yes

4 54 M 25 17 1 109 40 �72 Yes

5 58 F 34 23 5 122 43 �78 Yes

6 44 M 24 27 1 110 129 �92 Yes

7 45 F 26 1 0 113 32 �39 Yesa

8 44 M 31 1 4 141 0 5 No

9 52 M 29 12 14 137 �54 n/a No

10 51 M 24 22 21 130 �6 4 No

aPrimary snoring, snoring significantly improved after treatment.

Fig. 3. Significant correlations between upper airway volume change and mandibular angle (1); between change in upper airway volume and change in

AHI (2); between change in upper airway resistance and upper airway volume (3); between change in AHI and change in upper airway resistance (4).

J.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–3714 3711

had an average AHI change of 16.978.4. The non-responders had an average AHI change of �1.372.1.

A positive change or enlargement in UA volumecorrelates (r ¼ �0.70, p ¼ 0.02) with clinical improvement(Fig. 3). The responders had an average increase in UAvolume of 75.9766.2%. Simultaneously, no change or

even a decrease in UA volume was noticed in patientswithout clinical improvement with an average volumechange of �20.0729.6.The mean AM was 114.675.3 for the responders and

136.075.6 for the non-responders. A significant correla-tion is found between AM and change in UA volume

ARTICLE IN PRESSJ.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–37143712

(r ¼ �0.68, p ¼ 0.03) indicating that a larger AM leads to anegligible or even negative change in UA volume (Fig. 3).

A mean value of �65.0719.2 was found for the UAresistance change in the responders and 4.570.7 for thenon-responders. A significant correlation is obtainedbetween UA resistance and UA volume change(r ¼ �0.79, p ¼ 0.02) (Fig. 3). A decrease in AHI correlatesvery significantly to a decrease in airway resistance(r ¼ 0.92, p ¼ 0.001) (Fig. 3). For two patients the changein UA resistance is not included since the models showedan occlusion in one of the two states (with or withoutMAD). Subsequently, no flow simulation could be done.

Fig. 4 shows velocity contours and vectors inside the UAwithout and with MAD for a responder. The figureillustrates the velocity increase at the narrowest CSA.The vectors show the reduction in velocity and the

Fig. 4. Velocity contours inside upper airway of SBD patient (upper panel); ve

(lower panel).

corresponding increase in pressure at the narrowest CSAwhen the MAD is used.

4. Discussion

The precise mechanism of action of MADs is quitecomplex and not yet completely understood. Interindivi-dual variations in treatment success are obvious andimportant, while the treatment success is difficult topredict.Many papers assess the issue of possible predictors of

treatment success with MAD. Mehta et al. (2001) were ableto develop a predictive model based on a positivecorrelation between success with MAD and both neckcircumference and baseline AHI. Marklund et al. (1998)reported supine-dependent sleep apnea as the strongest

locity vectors inside the upper airway without (left) and with (right) MAD

ARTICLE IN PRESSJ.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–3714 3713

predictor of success with MAD. In a recent retrospectivestudy, the non-responders to the MAD treatment hadsignificantly larger oropharyngeal CSA based on cephalo-metric variables, and exhibited significantly higher weightgain compared to matched responders to MAD (Otsuka etal., 2006). In general, however, no consensus exist on auniversal marker for treatment success and more researchhas to be done (Cistulli and Gotsopoulos, 2004).

In our study, we have demonstrated that it was possibleto reliably predict treatment outcome in 10 SBD patientsusing functional imaging of the UA based on CFD. Twoparameters were shown to correlate significantly with theAHI: the UA volume and the UA resistance. Of these, theUA resistance provides the highest correlation coefficient.The AM was shown to correlate significantly with the UAvolume. Previous papers have shown a correlation betweenUA morphology and severity of sleep apnea (Schwab et al.,2003; Vos et al., 2007). From these studies it could beconcluded that airway dimensions can be an indication ofthe severity of OSAHS. Oral appliances such as MADs aimto correct the UA morphology in such a way that thesymptoms of OSAHS reduce. Our study confirms thefinding that whenever the UA morphology is adequatelycorrected (resulting in an enlargement of the UA volumeand a decrease of the UA resistance), the manifestation ofOSAHS is less severe.

The question remains why the UA volume tends toincrease in some patients and remains virtually unchangedin others. In this respect, the correlation between the AMand the change in UA volume, shown in Fig. 3 mightcontribute to a better understanding of the mode of actionof MADs. From this correlation, it appears that patientswith a larger AM had a smaller increase in UA volumeand, hence, exhibit an inferior response to the therapy. Thisrelation might be explained by considering the anatomy ofthe UA region. Enlargement of the upper airway can beattributed to the movement of the tongue base determinedby the genioglossus muscle (GG). The GG muscle isattached to the forward part of the mandibula via the spinamentalis. Therefore, this part of the mandibula inherentlydetermines the change in UA shape and volume. The use ofan MAD results in a downward (due to the rotation) and aforward (due to the advancement) movement of themandibula. Subsequently, the AM determines the truehorizontal advancement. Fig. 2 illustrates the difference intrue horizontal advancement for a responder and a non-responder. It can be seen that the responder shows agreater horizontal advancement and, hence, a larger UAvolume increase. The effective advancement is thereforedetermined by both the AM and the protrusion. In thepresent study, it appeared that for some patients thevertical movement of the mandibula induces a rotation ofthe tongue base resulting in a narrowing of the UA insteadof an enlargement. This can be seen in the non-responder inFig. 2 where the UA is obstructed with MAD.

The above observation shows that the mode of action ofa MAD is complicated by the vertical movement of the

mandibula due to the rotation. It can be expected that insome patients, an MAD will provide a global increase inUA volume, while locally the narrowest CSA decreases as aresult of the rotation of the tong base. In this respect, thedetermination of the UA resistance using CFD is essential,since this parameter will reflect local decreases in CSAcaused by the rotational movement of the tongue base.One could argue that a limitation of the method

presented is that the CT scans are taken during wakeful-ness, which will result in a different upper airway geometrythan during sleep. Although this is undoubtedly the case,the availability of a predictor for success that can beobtained by examining patients in awake state wouldgreatly contribute to the routine applicability of thetechnique. In addition, Schwab et al. (1993, 2003) haveshown that certain parameters such as enlargement of thesoft tissue measured during wakefulness can be regarded asrisk factors for sleep apnea. It is therefore reasonable toassume that certain parameters obtained during wakeful-ness can be predictors for treatment success.In summary, the aim of the project was to assess whether

the efficiency of a MAD could be predicted based onpatient specific anatomical features and CFD analyses. Theresults indicate that prediction of outcome in individualpatients is possible and that certain anatomical and flowcharacteristics can be considered as markers for success.This study demonstrates that, for the 10 patients con-sidered, it was possible to correctly predict the outcome ofMAD treatment based on the change in UA volume, thechange in airway resistance and the AM.The results of this proof of concept paper will be

implemented in our future protocols for MAD treatment.The predictive value will then be assessed based on a largernumber of subjects. It can be expected that this high-techUA model will find his way into clinical medicine to provehis value. The use of the model in clinical practice mightsignificantly increase the success rate with MAD or othertreatments for SBD, thereby also improving the cost-benefit of treatment of SBD in general.

Conflict of interest

This manuscript has been prepared according to allethical and scientific guidelines. No conflict of interestexisted during the course of this study and the preparationof the manuscript.

References

American Academy of Sleep Medicine Task Force, 1999. Sleep-related

breathing disorders in adults: recommendations for syndrome defini-

tion and measurement techniques in clinical research. pp. 667–689.

Boudewyns, A.N., De Backer, W.A., Van de Heyning, P.H., 2001. Pattern

of upper airway obstruction during sleep before and after uvulopala-

topharyngoplasty in patients with obstructive sleep apnea. Sleep

Medicine 2, 309–315.

Cistulli, P.A., Gotsopoulos, H., 2004. Single-night titration of oral

appliance therapy for obstructive sleep apnea: a step forward?

ARTICLE IN PRESSJ.W. De Backer et al. / Journal of Biomechanics 40 (2007) 3708–37143714

American Journal of Respiratory and Critical Care Medicine 170,

353–354.

Cistulli, P.A., Gotsopoulos, H., Marklund, M., Lowe, A.A., 2004.

Treatment of snoring and obstructive sleep apnea with mandibular

repositioning appliances. Sleep Medicine Review 8, 443–457.

De Backer, W., 2006. Obstructive sleep apnea–hypopnea syndrome—

definitions and pathophysiology. In: Randerath, W.J., Sanner, B.M.,

Somers, V.K. (Eds.), Sleep Apnea—Current Diagnosis and Treatment.

S. Karger AG, Basel, Switzerland, pp. 90–96.

Ferguson, K.A., Cartwright, R., Rogers, R., Schmidt-Nowara, W., 2006.

Oral appliances for snoring and obstructive sleep apnea: a review.

Sleep 29, 244–262.

Hoekema, A., Stegenga, B., De Bont, L.G., 2004. Efficacy and co-

morbidity of oral appliances in the treatment of obstructive sleep

apnea–hypopnea: a systematic review. Critical Reviews in Oral Biology

and Medicine 15, 137–155.

Huang, Y., White, D.P., Malhotra, A., 2005. The impact of anatomic

manipulations on pharyngeal collapse: results from a computational

model of the normal human upper airway. Chest 128, 1324–1330.

Kato, J., Isono, S., Tanaka, A., Watanabe, T., Araki, D., Tanzawa, H.,

Nishino, T., 2000. Dose-dependent effects of mandibular advancement

on pharyngeal mechanics and nocturnal oxygenation in patients with

sleep-disordered breathing. Chest 117, 1065–1072.

Kushida, C.A., Morgenthaler, T.I., Littner, M.R., Alessi, C.A., Bailey, D.,

Coleman, J., Friedman, L., Hirshkowitz, M., Kapen, S., Kramer, M.,

Lee-Chiong, T., Owens, J., Pancer, J.P., 2006. Practice parameters for

the treatment of snoring and obstructive sleep apnea with oral

appliances: an update for 2005. Sleep 29, 240–243.

Levy, P., Viot-Blanc, V., Pepin, J.L., 2006. Sleep disorders and their

classification—an overview. In: Randerath, W.J., Sanner, B.M.,

Somers, V.K. (Eds.), Sleep Apnea—Current Diagnosis and Treatment.

S. Karger AG, Basel, Switzerland, pp. 1–12.

Lim, J., Lasserson, T.J., Fleetham, J., Wright, J., 2006. Oral appliances for

obstructive sleep apnoea. Cochrane Database System Review

CD004435.

Marklund, M., 2006a. Oral appliances in the treatment of obstructive

sleep apnea and snoring. In: Randerath, W.J., Sanner, B.M., Somers,

V.K. (Eds.), Sleep Apnea. Karger, Basel, pp. 151–159.

Marklund, M., 2006b. Predictors of long-term orthodontic side effects

from mandibular advancement devices in patients with snoring and

obstructive sleep apnea. American Journal of Orthodontics and

Dentofacial Orthopedics 129, 214–221.

Marklund, M., Persson, M., Franklin, K.A., 1998. Treatment success with

a mandibular advancement device is related to supine-dependent sleep

apnea. Chest 114, 1630–1635.

Marklund, M., Stenlund, H., Franklin, K.A., 2004. Mandibular advance-

ment devices in 630 men and women with obstructive sleep apnea and

snoring: tolerability and predictors of treatment success. Chest 125,

1270–1278.

Mehta, A., Qian, J., Petocz, P., Darendeliler, M.A., Cistulli, P.A., 2001. A

randomized, controlled study of a mandibular advancement splint for

obstructive sleep apnea. American Journal of Respiratory and Critical

Care Medicine 163, 1457–1461.

Ng, A.T., Gotsopoulos, H., Qian, J., Cistulli, P.A., 2003. Effect of oral

appliance therapy on upper airway collapsibility in obstructive sleep

apnea. American Journal of Respiratory and Critical Care Medicine

168, 238–241.

Otsuka, R., Almeida, F.R., Lowe, A.A., Ryan, F., 2006. A comparison of

responders and nonresponders to oral appliance therapy for the

treatment of obstructive sleep apnea. American Journal of Orthodon-

tics and Dentofacial Orthopedics 129, 222–229.

Putz, R., 1994. Sobotta. Bohn Stafleu Van Loghum, Houtem.

Ryan, C.F., Love, L.L., Peat, D., Fleetham, J.A., Lowe, A.A., 1999.

Mandibular advancement oral appliance therapy for obstructive sleep

apnoea: effect on awake calibre of the velopharynx. Thorax 54,

972–977.

Schwab, R.J., Gefter, W.B., Hoffman, E.A., Gupta, K.B., Pack, A.I.,

1993. Dynamic upper airway imaging during awake respiration in

normal subjects and patients with sleep disordered breathing.

American Review of Respiratory Diseases 148, 1385–1400.

Schwab, R.J., Pasirstein, M., Pierson, R., Mackley, A., Hachadoorian, R.,

Arens, R., Maislin, G., Pack, A.I., 2003. Identification of upper airway

anatomic risk factors for obstructive sleep apnea with volumetric

magnetic resonance imaging. American Journal of Respiratory and

Critical Care Medicine 168, 522–530.

Smith, D.M., Stradling, J.R., 2002. Can mandibular advancement devices

be a satisfactory substitute for short term use in patients on nasal

continuous positive airway pressure? Thorax 57, 305–308.

Tegelberg, A., Walker-Engstrom, M.L., Vestling, O., Wilhelmsson, B.,

2003. Two different degrees of mandibular advancement with a dental

appliance in treatment of patients with mild to moderate obstructive

sleep apnea. Acta Odontologica Scandinavica 61, 356–362.

Tsuiki, S., Lowe, A.A., Almeida, F.R., Fleetham, J.A., 2004. Effects of an

anteriorly titrated mandibular position on awake airway and

obstructive sleep apnea severity. American Journal of Orthodontics

and Dentofacial Orthopedics 125, 548–555.

Tsuiki, S., Almeida, F.R., Lowe, A.A., Su, J., Fleetham, J.A., 2005. The

interaction between changes in upright mandibular position and supine

airway size in patients with obstructive sleep apnea. American Journal

of Orthodontics and Dentofacial Orthopedics 128, 504–512.

Vanderveken, O.M., Boudewyns, A.N., Braem, M.J., Okkerse, W.,

Verbraecken, J.A., Willemen, M., Wuyts, F.L., De Backer, W.A.,

Van de Heyning, P.H., 2004. Pilot study of a novel mandibular

advancement device for the control of snoring. Acta Otolaryngologica

124 (5), 628–633.

Vanderveken, O.M., Oostveen, E., Boudewyns, A.N., Verbraecken, J.A.,

Van de Heyning, P.H., De Backer, W.A., 2005. Quantification of

pharyngeal patency in patients with sleep-disordered breathing.

Journal of Otorhinolaryngol Related Spectrum 67, 168–179.

Vos, W., De Backer, J., Devolder, A., Vanderveken, O., Verhulst, S.,

Salgado, R., Germonpre, P., Partoens, B., Wuyts, F., Parizel, P., De

Backer, W., 2007. Correlation between severity of sleep apnea and

upper airway morphology based on advanced anatomical and

functional imaging. Journal of Biomechanics 40 (10), 2207–2213.

Walker-Engstrom, M.L., Ringqvist, I., Vestling, O., Wilhelmsson, B.,

Tegelberg, A., 2003. A prospective randomized study comparing two

different degrees of mandibular advancement with a dental appliance

in treatment of severe obstructive sleep apnea. Sleep Breath 7, 119–130.

Wilcox, D.C., 1998. Turbulence Modelling for CFD. California DCW

Industries Inc.