Aimah of Oliihgy. Rhiniil«gy & Uii-yngologv 1 l4[!2):94Q-957.© 2OO.'> Annals Publishing Company. All right.s reserved.

Acoustic Rhinometry: Accuracy and Ability to Detect Changes inPassage Area at Different Locations in the Nasal Cavity

Ozcan Cakmak. MD: Erkan Tarhan, MD; Mehmet Coskun. MD;Mehmet Cankurtaran, PhD; Hu.seyin Qelik, PhD

Objectives: To evaluate the accuracy of acoustic rhinometry (AR) measurements, and to assess how well AR detectsobstructions of various sizes at specific sites in the nasal cavity, we created a cast model from an adult cadaver nasalcavity.

Methods: The actual cross-sectional areas ofthe cast model nasal passage were determined hy computed tomographyand compared with the corresponding areas measured hy AR. To assess how nasal ohstruction affects the AR resulls. weplaced small wax spheres of different diameters at specific sites in the model (nasal valve, head ofthe inferior turbinate.head ofthe middle turhinate. middle ofthe middle turhinate. choana. and nasopharynx).

Results: The AR-derived cross-sectional areas in the first 6.5 cm of the cast model nasal cavity were very close to thecorresponding areas calculated from computed tomographie sections perpendieular to the presumed acoustic axis. How-ever. AR overestimated the passage areas at locations posterior to the 6.5-cm point. Acoustic rhinometry gave an accu-rate indication of the passage area of the nasal valve and its distance from the nostril. The nasal valve and the choanawere indicated by significant dips on the AR area-distance curve, whereas the curve was smooth throughout the regionthat included the head of the inferior turbinate. the head of the middle turhinate. the middle of the middle turbinate. andthe nasopharynx. In other words. AR did not discretely identify these latter sites. Acoustic rhinometry detected thedilTerent-si/.ed inserts (obstructions) more accurately at the nasal valve than at sites posterior to this location.

Conclusions: The results of the study show that AR is a valuable method tor assessing the anterior nasal cavity. Thistechnique is sensitive for detecting changes in passage area at the nasal valve region: however, the sensitivity is lower atsites posterior to this. The findings suggest that when there is substantial narrowing of the nasal valve. AR will notidentify an obstruction at any location posterior to the nasal valve. In such situations. AR measurements heyond theahnormal nasal valve may easily lead to misinterpretation ofthe patient's nasal anatomy or condition.

Key Words: acoustic rhinometry. cast model, computed tomography, curved acoustic axis, nasal cavity.

INTRODUCTION

For more thati a century, rhinologists have beensearchitig for a reliable way to evaluate the nasal air-way in a wide range of patients. In 1989. Hilberg etal' introduced acoustic rbinometry (AR) as an ob-jective method for examining the nasal cavity. Thistechnique is based on the principle that a sound pulsepropagating in the nasal cavity is reflected by localchanges in acoustic itnpedance. Acoustic rhinometryis a quick, painless, noninvasive method that can beusedtoestimatc the dimensions of nasal obstrnctions.evaluate nasal cavity geometry, monitor nasal disor-ders, and assess surgery results and response to tnedi-cal treatment. Because of these advantages, AR hasbeen widely accepted in a short period of time. How-ever, lack of standardization is one ofthe tnain prob-lems with this tnethod, and the role of AR in the clinic

is currently being evaluated.-In order to correctly interpret AR results, it is es-

sential to know how certain anatomic structures inthe nasal cavity are reflected on the AR area-distancecurve, and thus be able to identify their locations onthis curve. However, inconsisteticy in the literaturehas caused confusion. Different terms have been usedto denote the same anatomic structute in the nose,and different names have been proposed for corre-sponding cross-sectional areas on AR curves.-^"* Itshould also be noted that AR actually tneasiires cross-sectional area as a function of distance (from the nos-tril into the nose) along an estimated acoustic axisthat passes through the center of the curved nasalairway, as opposed to a straight litie parallel to thefloor of the nose. '" ' -This curved course of the na-sal cavity complicates the interpretation of AR data.

From the Dcparlnienls of Otorhinoliirynuology {Cakmak. Taihan) and Radiolojiy (Coskun). Baskenl t.lnivcrsity Faculty of Medicine,and ihe Department of Physics iCankiirtaran. ^'olikl. Hiteettepc University Faculty of Enginecting. Ankara. Turkey. This work waspartially supported by the BaskL'ni University Research huiid (proiect KA()4/(W).

Correspondenfe: Ozcaii Cakmak, MD, Baskent Universilesi Hastancsi, Kulak Burun Bogaz Anabilim Dali, Baheelievler, 06490,Ankara. Tiiikev.

949

950 Cakmak et al. Acoustic Rhinometrx

Fig 1. Three-dimensional image of luminai impressionof cadaver nasal cavity reconstructed by computed to-mography. Dotted tine represents estimated acoustic axis,

and raises questions about the validity of compari-sons between nasal cavity dimensions obtained byAR and those derived from cotnputed tomography(CT) or magtietic resonance imaging. Most publishedstudies have not considered this curved shape of thenasal cavity and acoustic pathway, and this neglectcould lead to misinterpretation of AR results.''•'-•'^Moreover, recent experimental and theoretical stud-ies of pipe models simulating nasal passage anatomybave detnonstrated that certain factors inherent to thephysics and algorithms used in AR litnit the accuracyof this method, and lead to systematic measurementerrors.'••'"-" Nevertheless, the results of these pipemodel studies still need to be confirmed with AR mea-surements of more realistic nasal cavity models.-'Finally, the literature contains very little informationabout the accuracy and sensitivity of AR measure-ments at different locations within the nasal cav-ity.'-----^Thus. there is also a need to examine howchanges in the dimensions of different parts of thenasal cavity influetice the accuracy of AR measure-ments.

The aims of this study were to evaluate the accu-racy of AR tneasuretiients and to assess how wellthis tnethod detects obstructions of various sizes atspecific sites in the nasal cavity. To carry out the in-vestigations, we used a cast model of an adult ca-daver nasal cavity. We measured the actual cross-sectional areas of the cast model using CT and thencompared these with AR-derived areas. Then weplaced small wax spheres of different diameters atspecific locations within the cast model. These sitescorresponded to the nasal valve, the head of the in-ferior turbinate. the head of the middle turbinate, themiddle of the tniddle turbinate. the choana. and thenasopharynx. We reviewed the previously proposeddefinitions for these anatomic structures in the nose.and attetnpted to identify their locations on the ARarea-distance curve. The acoustic properties of ourcast model are not identical to those of a vital nasal

Fig 2. Cast model of cadaver nasal cavity. Sites at whichspherical wax inserts were placed arc shown: 1 —nostril:2 — nasal valve: 3 — head of inferior turbinate: 4 —head of middle turbinate: 5 — middle of middle lurbinate:6 — choana; 7 — nasopharynx. Dotted line representsestimated acoustic axis,

cavity, but investigation with this type of nasal pas-sage model can provide useful information about thevalue of AR in the clinical setting.'"- '

MATERIALS AND METHODS

A cast tnodel of a human cadaver nasal cavity wastnade as follows. A luminai impression was preparedby injecting the right nasal passage of an adult ca-daver with silicotie (RTV-.'iO. Rhone-Poulenc. SaintFons. France). The tnaterial was allowed to set andwas then removed from the nose by dissection. Thisluminai impression (Fig I) was then used to cast anasal cavity model (Fig 2). The walls of the cast na-sal passage model were prepared from plaster mate-rial. The posterior portion of the cast model passage-way beyond the choana was rasped to enlarge thecontracted cadaver nasopharyngeal cavity to normaladult dimensions. To prevent acoustic leakage andrandom user-related errors, we tnolded the plastermaterial anterior to the cast model nostril to form acylindrical guide, thus ensuring that the nosepieceof the acoustic rhinometer fit tightly to the model(Fig 2).

Cakmak et al. Ac<>ti.';!ic Rhliumietrv 951

First., to evaluate the accuracy of AR measurementsat specific locations in the nasal cavity (see detailsbelow), we initially compared CT-derived and AR-derived area-distance curves for the cast model withno insert in place. In addition to examining for dif-ferences between these curves, we looked for changeswithin each curve (ie, wbether the area measuretnentswith either method distinguished the sites of interestin the study). The distance from the nostril (at 0.0cm) to each of the designated sites inside the pas-sage model (Xo) was measured along the estimatedacoustic axis (Fig 2). The distances recorded for thesites of tbe nasal valve, head of the inferior turbi-nate, head of the middle turbinate, middle of the mid-dle turbinate, choana, and nasopharynx were 2.0,2.8,4.0. 5.8, 8.8, and 10.0 cm, respectively. These wereall in excellent agreement with the corresponding dis-tances detemiined from CT of the luminal impres-sion.

Second, to assess AR detection of simulated nasaldisorders, we inserted wax spheres of various diam-eters at specific locations in the cast model. The 7sites of sphere placement chosen were based on pre-vious definitions'-*"^ of the locations of importantnasal cavity structures: the nostril and the 6 sites ofinterest inside the passage listed above. A set of 5spherical inserts (diameters 0.3, 0.5.0.7,0.9. and 1.1cm) was tested at each of these sites of interest in thetnodel. For each set of AR measurements, the insertswere placed in sequence (from smallest to largest) ateach designated site. To assess whether AR was ableto detect sitnulated nasal disorders at the 7 locations,we placed the 5 different inserts otie by one at eachsite and recorded the AR area-distance curve for eachinsert size (ie, a set of 5 AR area-distance curves persite). If there was a measurable difference within theset of areas recorded for any particular site, this wasconsidered successful detection of the insert at thatlocation.

A ttansient-signal acoustic rhinometer (Ecco Vi-sion, Hood Instruments. Petnbroke. Massachusetts)was used to perform the acoustic measurements. Theprocessed bandwidth for this rhinometer ranged from100 Hz to 10 kHz, and a lO-kH/. low-pass filter wasused to reduce the errors associated with cross modesin the actual nasal cavity. The same nosepiecc wasused to obtain all measurements presented in this ar-ticle. All AR measurements were repeated at least 5titnes to ensure that the results were reproducible.Data collected from the examinations were analyzedwith Origin software (version 7.0, Microcal SoftwareItic. Northampton. Massachusetts).

Actual cross-sectional areas for the first 8.8 cm oftbe cast model nasal passage (ie, from nostril to cho-ana) were determined from a CT scan ofthe luminal

impression ofthe cadaver nasal passage. The CT ex-aminations were performed with a multislice scan-ner (Somatom Sensation 16. Siemens. Erlangen, Ger-many) with tube voltage of 120 kV and current of240 mA. The window width was 4,000 Hounsfieldunits, and the window level was centeted at 600Hounsfield units. Axial CT slices parallel to the longaxis of the luminal impression of the cadaver nasalcavity were obtained with 0.75-mtn coilimation, 2-mm slice thickness, and 5-mm table feet, and theseimages were reconstructed with l-tnm intervals bymeans of a soft tissue algorithm. Frotn the recon-structed axial slices, a 3-dimensional (3-D) shadedsurface display image ofthe luminal impression wasobtained (Fig 1).

As described in previous studies,' •" ' - the concep-tualized acoustic axis for the AR determinations fol-lows an estimated line that passes through the centerofthe nasal passage, and AR-derived distances fromthe nostril to sites along this line would be the sameirrespective of passageway shape. To determine tbecross-sectional areas perpendicular to the estimatedacoustic axis, we divided the acoustic pathway ofthe luminal impression of the cadaver nasal passageinto 2 segments and drew each manually in a 3-Dreconstructed image (Fig 1). The first segment wasin the anterior nasal cavity and fortned a quarter-cir-cle; it extended from the center of the nostril open-ing and ended at the head of the inferior turbinate.The second segment ofthe acoustic axis was drawnas a straight line parallel to the long axis ofthe infe-rior turbinate: it extended from the end point of thefirst axis segtnent to the choana. The length of eachsegment of the acoustic axis was measured on CT,and 30 slices perpendicular to the acoustic axis wereobtaitied for each segtnent (ie, 60 cross-sectional ar-eas recorded for the first 8.8 cm of the model nasalpas.sage). A plot ofthe slice area versus the distancealong the estimated acoustic axis yielded the CT-de-rived area-distance curve.

RESULTS

Figure 3 shows a graphic comparison of the ac-tual cross-sectional areas of the cast model (deter-mined from CT ofthe luminal impression) and thecorresponding AR-derived cross-sectional areas. TheAR data were in good agreement with the CT datathrough the first 6.5 cm of the cast model passage.For the nasal valve (the narrowest part of the nasalpassage), the AR and CT evaluations yielded altnostidentical data for cross-sectional area (passage area:1.08 cm-) and location relative to the nostril (2.0 cm).The results indicate that in cases in which the pas-sage area of the nasal valve is in the normal adultrange. AR measurements ofthe anterior nasal cavity

952 Cakmak el a!. Aiou.stic Rhiniimefr

< 2

—• —CT Area—o—AR Arou

3 4 5 6 7

Distance (cm)10 II 12

Fig 3. Comparison of eross-sectional areas of east nasaleavity model derived from aeoustie rhinometry (AR: openeireles) and aetual eross-sectional areas caleuiated fromeoniputed tomography (CT) ol' luminal impression(seanning done perpendicular to aeoustie axis; solidcircles). Numbers 1 to 6 on vctiical dotted lines representactual sites of nostril, nasal valve, head of inferiorturhinate. head of middle turbinate. middle of middleturhinate. and ehoana, respectively.

are preci.se. In contrast, at distances farther than 7.0cm inside the nostril, the AR-derived cross-sectionalareas were overestimated as compared to those ob-tained from CT. The degree of this area overestima-tion iticreased with distance, and was extremely high(about 70%) at the location of the choana (8.8 cm).

Careful inspection of Fig 3 reveals that the AR-derived area-distance curve was fairly smooth throughthe part of the nasal passage that includes the headofthe inferior turbinate (site 3 in Fig 3), the head ofthe middle turbinate (site 4). and the middle of themiddle turbinate (site 5). In other words. AR did notdistinctly identify any of these structures. For thehead ofthe inferior turbinate, the same held true onthe CT-derived curve. At the location ofthe head ofthe middle turbinate, there was a small but measur-able dip in the CT curve, and a very slight decreasein the slope of the AR curve. At the location of themiddle of the middle turbinate, slight decrea.ses inslope were noted on both curves. On both the CTand AR curves, the choana (site 6 in Fig 3) was re-flected as deep minima; however, as noted. AR over-estimated the cross-sectional area at the choana byabout 70%. These findings suggest that even whenthe nasal valve passage area is in the normal adultrange, AR only discretely identified the locations ofthe nasal valve and the choana.

Figure 4 illustrates the set of AR-derived area-dis-tance curves that correspond to the cast model withan unobsttucted nostril and with spherical insetts of4 different diameters (d) placed in the nostril. TheAR-measured cross-sectional area at the site of the

4 5 6 7

Distance (cm)

Fig 4. Aeoustie rhinometry area-distance curves for casttnodel with no insert (ohslruction) and with in.serts ofdiameter d placed at entrance of nostril. Different sym-hols refer to data sets ohiained with inserts of differentdiameier.

nostril decreased from 0.9 to 0.4 cm- as the diameterofthe insert increased from 0.3 to 0.9 cm. These re-sults confinn that the first tninimum on the AR area-distance curve (site I in Fig 3) corresponds to thejunction between the end of the nosepiece of theacoustic rhinotneter and the nostril.

In the next sets of experiments, we examined thecast tnodel with spherical wax inserts of diatneter dplaced at specific distances (Xo) from the nostril. Asnoted, this simulated a di.sorder at different sites with-in the nasal passage. Figure 5 shows sets of AR area-distance curves for the cast model with inserts lo-cated at Xo of 2,0 (nasal valve), 2.8 (head of inferiorturbinate), 4.0 (head of tniddle turbinate). 5.8 (middleof middle turbinate). S,S (choana). and 10.0 cm (na-sopharynx). Each set contains 6 curves: one corre-sponding to no insert and one for each ofthe 5 insertsizes. One feature comtnon to all ofthe data sets pre-sented in Fig 5 is that regardless of insert location,the AR-measured cross-sectional areas anterior to theobstruction were similar and almost completely in-dependent of the size of the obstruction. The resultssuggest that in a cast model, regardless ofthe degreeof obstruction, AR yields similar cross-sectional ar-eas from the nostril to approximately 1.0 cm beforethe obstruction. The data also show that beyond alarge obstruction in the nasal passage, AR underesti-mates cross-sectional area. The degree of area under-estimation depends on both the diameter and the lo-cation ofthe insert.

The same group of graphs (Fig 5) reveals that ARwas more sensitive in detecting diffetent-sizcd spher-ical inserts placed at the nasal valve than in detect-ing inserts at tnore posterior sites. The AR area-dis-tance curve obtained with the 0.3 cm diameter insert

rP 3E

< 2

Cakmak el ai Acoiislic Rhinometry

5

953

— • —WJihoul iiisen

—o —d = 0.3 cm-A- d - 0.5 cm—zi— d = 0-7 cm— • — d - 0 . 9 cm- o - d 1.1 cm

0 1 2 3 4 5 6 7 X

Distance (cm)I (I 11 12

-1 0 I 2 3 4 5 () 7

Distance (cm)10 I I 12

6 7 a 9 Id 11 12

4 -

B

0 1 2 1 4 5 6 7 8 9 10 I I 12

0 1 2 3 4 5 6 7 8 9 1 0

Distance (cm)12

Distance (cm)4 S 6

Distance (cm)

Fig 5. Acoustic rhinometry area-distance curves for cast model with no insert (obstruction) and with inserts of diameter d placedinside casi model at A) nasal valve. B) head of inlerior turbinate, C) head of middle turbinate. D) middle of middle lurbinate. E)ehoana. and F) nasopharynx. Different symbols refer lo daia seis ohiained wiih inserisof differeni diameier. Xo — distance fromnostril.

placed at the nasal valve of the model was almostidentical to that for the tnodel without an insert (Fig5A). This is because the effective cross-sectional areaof this particular insert was very small (0.07 cm-)

and therefore did not significantly reduce pa.s.sagearea at this location. However, when inserts of 0.5cm diameter or larger were placed in the nasal valve,the AR-derived area of the nasal valve decreased sub-

954 Cakmak et al. Acoustic Rhinomelty

^ " 3

-°—Withoui insert-•—With insert (ll 0.7 tml a! (hi; nusal valve-•—Case I-•—Case II-A—Case III

2 3 4 .S 6 7 S

Distance (cm)111 II 12

Fig 6. Acoustic rhinometry area-distance curves obtainedfor cast model with no insert, with inserl at nasal valve.and with combinations of 2 (case I and case II) and 3(case III) simullaneous obstructions. Case I — I insertplaced at nasal valve and another at head of middle tur-binate; Case II — I insert placed at nasal valve and an-other at choana: Case III — inserts placed at nasal valve,head of middle turbinate. and choana, Dianielers of insertsplaced at nasiil valve, head of middle turbinate, and eho-ana were 0.7, 0,9, and 0.9 eni, respectively.

stantially. and the passage area values beyond theinsert were consistently underestimated (Fig 5A).

When inserts were placed in locations posteriorto the nasal valve, the sensitivity of AR in detectingchanges in passage area decreased. Whereas AR de-tected the 0.5 cm diameter insert at the head of theinferior turbinate, the smallest inserts that could bedetected at the head of the middle turbinate. the mid-dle of the middle turbinate. and the choana were the0,7 cm. 0.9 cm. and 0.9 cm sizes, respectively (Fig5B-E). However, even when the largest insert (1.1cm diameter) was placed in the nasopharynx, therewas no appreciable change in cross-sectional areameasured by AR (Fig 5F). In other words, it appearsthat a disorder in the nasal valve region is likely tobe detected by AR and will have a major influenceon the results of an AR examination. However, dis-orders beyond the nasal valve are not likely to bedetected by AR.

We also recorded AR measurements for the castmodel with 2 or 3 inserts simultaneously placed atselected sites in the nasal pas.sage. In case I. an in-sert was placed at the nasal valve and another wasplaced at the head of the middle turbinate. In case 11.an insett was placed at the nasal valve and anotherwas placed at the choana. In case III. inserts wereplaced at the nasal valve, the head of the middle tur-binate, and the choana. The diameters of the insertsplaced at the nasal valve, head of the middle turbi-nate, and choana were 0.7 cm. 0.9 cm. and 0.9 cm.respectively. Figure 6 shows the AR area-distance

curves for these 3 cases, for the cast model withoutan insert, and for the cast model with a 0.7 cm di-ameter insert placed at the nasal valve. In all 3 casesthe insert (obstruction) at the nasal valve was suc-cessfully detected by AR. However, when there wasaninsertat this site, the second insert (cases ! and II)or the second and third inserts (case III) at more pos-terior sites were not identifiable on the AR area-dis-tance curve. In summary, the findings from this partof the study indicate that when the nasal valve re-gion is significantly narrowed. AR will not detectdisorders at sites beyond this region.

DISCUSSIONNone of the research done to date has established

how well AR-measured cross-sectional areas corre-late with actual areas in certain regions of the nasalcavity. Use of different terminology for a given ana-tomic structure in the nose causes confusion whenone is locating its site on an AR area-distance curve.The rhinology literature has introduced various namesfor key anatomic sites in the nose. For example, theterm "ostium intemum" has been proposed as a coun-terpart to "ostium externum" for denoting the nostrilor naris. Previously, Huizing^ discussed the nasal no-menclature that should be abandoned and replacedby anatomically, linguistically, and surgically cor-rect terms. He suggested using the term "nasal valvearea" for the narrowest region of the breathing path-way instead of "isthmus nasi" or "ostium internum."Kasperbauer and Kern̂ "̂ considered the head of theinferior turbinate to be a component of the nasal valvearea. Acoustic rhinometric examination of the ante-rior nasal cavity of normal human subjects has shown2 notches (or dips) on the AR area-distance curvethat indicate sites of very small cross-sectional area.Lenders and Pirsig"̂ described the first such low pointon the AR curve as the "l-notch" (I = isthmus) andnoted that this corresponded to the functional isth-mus nasi (nasal valve). They identified the secondlow point as the "C-notch" (C = concha) and notedthat this corresponded to the head of the inferior con-cha. Lenders and Pirsig found that in AR of controlsubjects, the I-notch consistently indicated the over-all smallest cross-sectional area. In contrast, they ob-served that in patients with turbinate hypertrophy (dueto allergic or vasomotor rhinitis), the overall small-est cross-sectional area was the C-notch. In line withthis observation, they referred to the 2-notch sectionof an AR curve in which the lowest notch correspondsto the isthmus nasi as a "climbing-W" and to the op-posite situation (lowest notch denoting the head ofthe inferior turbinate) as a "descending-W," The sameauthors rept)rted that for patients with turbinate hyper-trophy, the AR trace changes from a descending-W(characteristic of rhinitis and turbinate hypertrophy)

Cakmak et al. Acoustic Rhinometry 955

before decongestion to a climbing-W (more charac-teristic of nortnal subjects) after decongestion. Ac-cording to Lenders and Pirsig, these 2 AR curves wereidentical at the isthmus nasi, reflecting the absenceof nasal mucosa in this anatomic location. They notedthat the 2 AR curves separated at a point correspond-ing to the anterior inferior turbinate because of thepresence of congested nasal mucosa starting at thissite.

In AR of 21 patients with septal deviation, Grymeret al-* identified the narrowest part of the nose as theisthmus nasi and noted that this structure was locatedslightly more posteriorly than the anatotnic ostiuminternum. According to these authors, the isthmusnasi had the overall stnallest cross-sectional area be-fore decongestion and the ostium internum had theoverall smalle.st cross-sectional area after deconges-tion. Inalaterstudy of healthy subjects with no grossnasal disorders, Grymer et al-̂ noted that the lowestpoint on the AR curve (site of smallest cross-sectionalarea) was the head of the inferior turbinate beforedecongestion and that the lowest point shifted to thesite reflecting the isthmus nasi after decongestion.

Hatnilton et ai'' reported that the first low point(minitnum) on the AR curve always corresponded tothe same distance from the start ofthe acoustic path-way and also noted that its location was not affectedby acoustic leakage through the gap between thenosepiece and the nose. They attributed the first min-imum to the junction between the nosepiece and thenostril, as opposed to correlating it with nasal anat-omy alone. Tomkinson and Eccles'" also interpretedthe first low point, or tninimum (MI), as the junctionof the nosepiece with the nasal aperture, since theyfound that its location on the AR curve depended onnosepiece length. These authors concluded that thefirst minimum on an AR curve did not reflect ananatomic entity {and should be strictly interpretedas corresponding to the end of the nosepiece) andthat the second tninitnum (M2) represented the nasalvalve area. The same article noted that the inferiorturbinate might contribute to the magnitude of thesecond minimum.

The results from the present study of variationson a cast tnodel ofthe nasal cavity confirm the inter-pretations in 2 previous reports'̂ "*: the first minimumon the AR area-distance curve corresponds to thejunction between the end of the nosepiece and thecast model nostril, and the second corresponds to thenasal valve. A similar interpretation may apply tothe actual nasal cavity, although the acoustic prop-erties of the cast model are not identical to those of avital nasal cavity, and the geometry ofthe actual nasalcavity is somewhat tnore complicated than the model

we investigated. However, Hilberg et aP-̂ argued thatthe complex geometry ofthe actual nasal cavity doesnot significantly affect AR measurements.

In addition, our present study showed that anatom-ic structures posterior to the cast model nasal valve(ie, the head of the inferior turbinate, the head of themiddle turbinate, the middle of the middle turbinate,and the nasopharynx) were not distinguishable onthe AR curve. Possible explanations for this resulttnay be the limited spatial resolution of AR and thefact that AR is not able to reflect sharp changes incross-sectional area throughout the course ofthe nasalpassage.-" Furthermore, previous model studies''̂ "'̂ •-'demonstrated that the accuracy of AR measurementsdiminished at locations beyond a significant constric-tion at the nasal valve. These findings suggest thatdisorders that narrow the anterior nasal passage, suchas septal deviations, polyps, tumors, webs, strictures,or the nasal valve of a pediatric patient, tnay signif-icantly affect the AR measurements when equipmentintended for adults is used.'"^ '^•-' Therefore, for pa-tients with such conditions, the value of AR for eval-uating the entire nasal cavity is limited.

In a clinical study that involved magnetic reso-nance imaging, Corey et al̂ described valleys (ordips) on the AR curve that represented reductions incross-sectional area at specific distances from the in-terface, each corresponding to defined anatomic struc-tures in the nose. The authors suggested that the first,second, and third valleys (with respective cross-sec-tional areas CS A1, CSA2. and CS A3) on the AR curverepresented the nasal valve, the anterior end of theinferior turbinate, and the anterior end ofthe middleturbinate. respectively. However, in that study, themagnetic resonance images were taken perpendicularto the floor ofthe nose. Later, Corey et al̂ *' comparedAR with rigid endoscopy for assessing landmarks inthe nasal cavity. In the examinations, they touchedthe tip ofthe endoscope to a landmark and then mea-sured the distance from the endoscope tip to the mark-ing. With this method, they found that CSA2 andCSA3 corresponded to the anterior portions of theinferior and middle turbinates. respectively, Coreyet al-f" suggested that AR gives a good indication forcross-sectional areas CSA2 and CSA3, but does notclearly delineate the anatomic correlate of CSAI (thenasal valve). However, in the original study that de-scribed the AR technique, Hilberg et al' emphasizedthat the major probletn with comparing AR findingsto those from imaging techniques is that it is difficultto obtain data exactly in the same plane. Later, usinga plastic nose model produced by stereolithographictechniques from a 3-D tnagnetic resonance imagingscan, Djupesland and Rotnes" found good agreementbetween the nasal cavity volumes detertnined from

956 Cakmak et al. Acoustic Rhinometrv

AR-derived dimensions and magnetic resonance im-aging when the magnetic resonance cross sectionswere obtained perpendicular to the presumed courseot the sound pathway. More recently, Cakmak et al'^showed that when AR data for the nasal valve regionare to be compared with calculations based on anyimaging technique, the imaging cross sections mustbe taken perpendicular to the curved acoustic axis.They found a significant correlation between AR dataand data from CT images obtained perpendicular tothe curved acoustic pathway; however, there was nocorrelation between AR data and data from CT imag-ing done perpendicular to the nasal floor. Our presentfindings confirm the results of previous studies"'-that emphasize the importance of the curved acousticaxis: that is. distances on AR curves correspond ac-curately to distances within the nasal passage as mea-sured along the estimated acoustic pathway. Our re-sults also .show that it is not appropriate to make crit-ical statements about the validity of AR on the basisof CT or magnetic resonance imaging data from dif-ferent imaging axes. Previous studies not taking intoaccount the curved shape of the nasal cavity couldlead to misinterpretation of AR data.

Various imaging methods have been used to testthe accuracy of AR in clinical trials with healthy hu-man subjects. These have revealed significant corre-lations between the cross-sectional areas obtained byimaging techniques and the cross-sectional areas ob-tained by AR, with particularly high agreement inthe anterior nasal cavity.''-'-''^-^-^ However, with theexception of studies by Terheyden et a!-̂ and Cakmaket al,'^ in all previous investigations of living sub-jects the CT and magnetic resonance images weretaken perpendicular to the floor of the nose.''-'^'^Moreover, some investigations used the distancesfrom the anterior nasal spine, whereas others usedthe tip of the nose as the reference point.̂ •'-̂ •''̂ Clearly,it is difficult to make definitive .statements about thevalidity of AR on the basis of data trom different im-aging techniques, different imaging axes, and poten-tially different sites in the nasal passage due to differ-ent reference points.

Similar to studies in which imaging methods havebeen used in healthy human subjects, model studieshave shown that AR is valuable for measuring nasalvalve passage area if this area is within the normaladult range.'•''^' However, pipe model studies haveshown that certain factors inherent to the physics andalgorithms used in AR limit the accuracy of this meth-od and lead to systematic measurement errors.'*'"-"These studies suggest that the accuracy of AR mea-surements diminished at locations beyond a signifi-cant constriction at the nasal valve, and also at regionsposterior to a paranasal sinus ostium, depending on

the ostium size and/or the sinus volume. When thepassage area of the nasal valve was small. AR-derivedcross-sectional areas beyond the constriction wereconsistently underestimated and the correspondingarea-distance curves showed pronounced oscillations.Previous model studies also demonstrated that para-nasal sinuses led to area overestimation and oscilla-tion of the AR curve posterior to the sinus ostiumwhen the ostium was relatively large. 11 • i'̂ -1-7-29 ̂ j-ggunderestimation or overestimation. as well as oscilla-tions of the AR curve (the physical reasons for whichhave been discussed'*^-'), may easily lead to misin-terpretation of a patient's nasal anatomy or condition.To avoid such misinterpretations, AR should be per-formed in combination with other methods such asanterior rhinoscopy and endoscopy.

Acoustic rhinometry has been widely used in clin-ical trials, but very little is known about how changesin the anatomy of the nasal cavity affect the accuracyof AR measurements. Several authors have investi-gated the sensitivity of AR in realistic nose modelsand in the living human nasal cavity by placing insertsat various sites in the nasal passage.'—-^ Hilberg etal' placed 0.3 cm spheres at different locations in acast model and found that AR was able to positivelyidentify the presence or absence of these objects. Sim-ilarly, Lenders et al--' in.serted small pieces of wax(volumes, O.I to 0.6 cm-̂ ) at different locations incast models. They found that a wax piece 0.3 cm-̂ involume placed in the middle meatus was detectedby AR, but observed no changes in the AR curvewhen the same insert was placed in the posterior nasalcavity. Fisher et al-- placed 0.3, 0.3, and 0.7 cm di-ameter spheres into either the middle meatus or thenasal valve of living subjects to assess how well ARwould detect these objects. The smallest spheres (0.3cm diameter) were detected in only a minority ofcase.s. those 0.5 cm in diameter were detected in mostcases, and those 0.7 cm in diameter were identifiedin the large majority of cases. Although detection rateswere similar for both locations, the authors observedgreater changes in area and volume for the 0.3 and0.3 cm spheres placed in the middle meatus than fortho.se placed in the nasal valve.^^ However, in thepresent study we observed the opposite: AR was moresensitive in detecting different-sized spherical insertsin the nasal valve than in detecting them at all loca-tions posterior to the nasal valve.

CONCLUSIONS

The results of the present study show that whenthe passage area of the nasal valve is within the nor-mal adult range, AR measures cross-sectional areasin the anterior nasal cavity with high precision, where-as cross-sectional areas at posterior locations are

Cakmak et ai. Acoustic Rhinometrv 957

overestimated. The first minimum on the AR area-distance curve represents the junction between theend ofthe nosepiece and the nostril, and the secondone corresponds to the nasal valve. Acoustic rhinom-etry accurately measures both the passage area andthe location of the nasal valve. The nasal valve andchoana are identified by pronounced dips on the ARcurve, whereas the head ofthe inferior turbinate, thehead ofthe middle turbinate. the middle ofthe tniddleturbinate, and the tiasopharynx are not distinguish-

able on the curve. The reasons for this lack are thelimited spatial resolution of AR and the inability ofAR to reproduce sharp changes in cross-sectionalarea throughout the course ofthe nasal passage. Theresults also .show that AR is sensitive in detectingchanges in passage area at the nasal valve, but thatits sensitivity is much lower at posterior sites. Whenthe nasal valve region is significantly narrowed, ARdoes not detect obstructions at sites beyond this re-gion.

Acknowledgments; We thank Dr Ulkeiii Cilasun and Dr Aktaii Cclik from Baskenl University Faculty of Dentistry for Lhcir help inpreparing the cast model used in the sludy.

REFERENCES

1. Hilberg O, Jackson AC. Switt DL. Pederscn OF. Acousticrhinometry: evaluation of nasal caviiy geometry by acousticretleclion. J AppI Physiol 1989:66:295-303.

2. Hilberg O. Pedersen OF. Acoustic rhinometry: recom-mendations for technical specifications and standard operatingprocedures. Rhinol Suppl 2U00( suppl I6):3-I7, lErraluniin Rhi-nology20()l;39:119.|

3. Grymer LR Hilberg O. Elbrond O, Pedersen OF. Acousticrhinometry: evaluation ofthe nasal cavity with septal deviations,before and after septoplasty. Laryngoscope 1989:99:1180-7.

4. Lenders H. Pirsjg W. Diagnostic value of acoustic rhi-nometry: paiients with allergic and vasomotor rhinitis comparedwith normal controls, Rhinology 1990:28:5-16.

5. Grymer LF, Hilberg O, Pedersen OF. Rasmussen TR.Acoustic rhinometry: values from adults with subjective normalnasal patency. Rhinology 1991:29:3.^-47,

6. Riechelmann H, Rheinheimer MC, Wolfensberger M.Acoustic rhin{)mctry in pre-school children. Clin OtolaryngolAllied Sci l993:IH:272-7.

7. CoreyJP. GungorA. Nelson R. FredbergJ. Lai V. A com-parison of the nasal cross-sectional areas and volumes obtainedwith acoustic rhinometry and magnetic resonance imaging. Oto-laryngol Head Neck Surg 1997:117:349-54.

8. Huizing EH. Incorrect terminology in nasal anatomy andsurgery, suggestions for improvement. Rhinology 2003;41:129-33.

9. Hamilton JW. McRae RD. Jones AS, The magnitude ofrandom errors in acoustic rhinometry and re-interpretation ofthe acoustic profile. Clin Otolaryngol Allied Sci 1997:22:408-13.

10. Tomkinson A. Eceles R. Acoustic rhinometry: an expla-nation of some common artefacts associaled with nasaldecon-gestion. Clin Otolaryngol Allied Sci 1998:23:20-6.

11. Djupe.sland PG, Rotnes JS. Accuracy of acoustic rhmom-etry. Rbinology 2001:39:23-7.

12. Cakmak O. Coskun M. Celik H. Buyuklu F. O/luogluLN. Value ot acoustic rhinometry for [neasuring nasal valvearea. Laryngoscope 2003:113:295-302.

13. Min YG, Jang YJ, Measurements of cross-sectional areaof the nasal cavity by acoustic rhinometry und CT scanning.Laryngoscope 1995:105:757-9.

14. Gilain L.Coste A, Ricolfi F. et al. Nasal cavity geometrymeasured by acoustic rhinometry and computed tomography.Arch Ololaryngol Head Neck Surg 1997:123:401-5.

15. Buenting JE, Dalston RM. Smith TL, Drake AF. Artifactsassociated with acoustic rhinometric assessment of infants and

young children: a model study. J AppI Physiol 1994:77:2558-63.

16. Hamilton JW. Cook JA. Phillips DB. Jones AS. Limita-tions of acoustic rhinometry determined by a simple model. AetaOtolaryngol (Stockh) 1995:115:811-4.

17. Cakmak O. Celik H. ErginT. Sennaroglu L. Accuracy ofacoustic rhinometry measurements. Laryngoscope 2001:111:587-94, lErratum in Laryngoscope 2001:111:1117,)

18. Cankurtaran M. Celik H. Cakmak O. Ozluoglu LN, Ef-fects or the nasal valve on acoustic rhinometry measurements:a model study. J AppI Physiol 2003:94:2166-72.

19. Cakmak O. Celik H. Cankurtaran M. Buyuklu F. Ozgir-ginN. Ozluoglu LN. Effects of paranasal sinus ostia and volumeon acoustic rhinometry measurements: a model study, J AppIPhysiol 2O03;94:1527-35.

20. Celik H. Cankurtaran M. Cakmak O. Acoustic rhinometrymeasurements in stepped-tube models of the nasal cavity. PhysMed Biol 2004:49:371-86.

2L Cakmak O. Celik H. Cankurtaran M, Ozluoglu LN. Ef-fects of anatomieal variations of the nasal cavity on acousticrhinometry measurements: a model study. Am J Rhinol 2005:19:262-8.

22. Fisher EW. Daly NJ. Morris DP. Lund VJ. Experimentalstudies of the resolution of acoustic rhinometry in vivo. ActaOtolaryngol (Stoekh) 1994:114:647-50.

23. Lenders H. Pent/ S. Brunner M. Pirsig W. Follow-up ofpatient-s with inverted papilloma of the nasal cavities: computertomography, video-endoseopy, aeoustie rhinometry'? Rhinology1994;32:167-72.

24. Kasperbauer JL. Kern EB. Nasal valve physiology. Im-plications in nasal valve surgery. Otolaryngol Clin North Am1987:20:699-719.

25. Hilberg O. LyholmB.Michelsen A. Pedersen OF. Jacob-sen O. Acoustic reflections during rhinometry: spatial resolutionand sound lo.ss. J AppI Physiol 1998:84:1030-9,

26. Corey JR Nalbone VP, Ng BA. Anatomic correlates ofacoustic rhinometry as measured by rigid nasal endoscopy. Oto-laryngol Head Neck Surg 1999:121:572-6,

27. Terheyden H, Maune S. Mertcns J. Hilberg O. Acousticrhinometry: validation by ihree-dimensionally reconstructedcomputer tomographie scans. J AppI Physiol 2000;89:1013-21.

28. Hilberg O. Pedersen OF. Acoustic rhinometry: influenceof paranasal sinuses. J AppI Physiol 1996:80:1589-94,

29. MIynski R.GriJtzenmacherS. Lang C.MIynskiG. Acous-tic rhinometry and paranasal eavities: a systematic study in boxmodels. Laryngoscope 2003:113:290-4.