Accuracy of angular measurements and assessment ofdistortion in the mandibular third molar region onpanoramic radiographsRaahib Dudhia, BDSc, DClinDent,a Paul A. Monsour, BDSc, MdSc, PhD, FICD,b

Neil W. Savage, BDSc, MdSc, PhD, FFOP(RCPA), FICD,b and Richard J. Wilson, BSc, PhD,c

Brisbane, Queensland, AustraliaQUEENSLAND X-RAY AND THE UNIVERSITY OF QUEENSLAND

Objective. The objective of this study was to examine distortion on panoramic radiographs by assessing the accuracy ofangular measurements in the mandibular third molar region on panoramic radiographs using computed tomography (CT).Methods. Images from patients referred to a radiology practice for low-dose CT scans of their lower third molars wereused in this study. The angle between the long axes of the second and third molars was assessed on panoramicradiographs and the corresponding CT images. Apparent tilting of the tooth across the arch on the panoramic imageand the bucco-lingual inclination on CT were also recorded and compared.Results. Eighty-eight patients had 163 mandibular third molars. The mean difference between the panoramicmeasurements and the CT measurements was �1.44 degrees, indicating that on the panoramic radiograph the thirdmolar appeared less mesially inclined on average. This pattern was present in most cases, but in 64 cases the thirdmolar appeared more mesially inclined on the panoramic radiograph. It was not possible to predict the direction ofthe discrepancy based on the radiographic findings. The discrepancy between the 2 measurements was greater if thethird molar appeared tilted across the arch on the panoramic image, and tilted teeth showed a higher bucco-lingualinclination on reformatted CT images.Conclusions. Distortions inherent in panoramic imaging because of projection geometry produce discrepancies in theangular measurements in the mandibular third molar regions on panoramic radiographs. Interpretation of third molarangulation from panoramic radiographs is often unreliable and may not accurately reflect the true orientation of thetooth. Apparent tilting of the tooth across the arch on the panoramic radiograph exacerbates the problem and appearsto correlate with the bucco-lingual inclination of the third molar as visualized on reformatted CT images. (Oral Surg

Oral Med Oral Pathol Oral Radiol Endod 2011;111:508-516)

Third molar development and eruption is closely mon-itored by dental practitioners. Assessment of third mo-lars is often required, as third molars are reportedly themost frequently impacted teeth.1-3 Panoramic radio-graphs are widely used in dentistry, as they permitvisualization of the dental arches and associated struc-tures with a relatively low radiation exposure. How-ever, owing to tomographic movement and the projec-tion geometry, distortion is introduced into the resultingimage, and this may affect interpretation of the radio-graph. Furthermore, because of superimpositions and

aDento-maxillofacial Radiologist, Queensland X-Ray, Brisbane,Queensland, Australia.bProfessor, The University of Queensland, School of Dentistry, Fac-ulty of Health Sciences, Brisbane, Queensland, Australia.cSenior Lecturer, The University of Queensland, Department ofMathematics, School of Physical Sciences, Brisbane, Queensland,Australia.Received for publication Aug 9, 2010; returned for revision Dec 5,2010; accepted for publication Dec 8, 2010.1079-2104/$ - see front matter© 2011 Mosby, Inc. All rights reserved.

doi:10.1016/j.tripleo.2010.12.005

508

the resulting 2-dimensional image, visualized anatomyon the panoramic radiograph may not accurately rep-resent the true anatomy of the patient. The presence ofdistortion in the horizontal plane is recognized, but quantifi-cation of such distortion is not readily facilitated.

Panoramic radiographs have been suggested for rou-tine third molar examination because of frequent anom-alies of third molar development, morphology, posi-tion, and eruption.4-6 When present, eruption of thirdmolars typically occurs between the ages of 14 and 24years.7 Third molar impaction is reported to occur in18% to 32% of the population,8 but impaction rates ofup to 40% have been noted.7 It has been reported thatsurgical removal of mandibular third molars is the mostfrequently performed oral surgical procedure.9 Giventhe postoperative morbidity that can occur followingsurgical removal of third molars, some authors haveattempted to predict the likelihood of third molar erup-tion to ensure appropriate management and minimizeunnecessary surgical removal.

Angulation of the third molar and the available space

in the third molar region reportedly influence the like-

OOOOEVolume 111, Number 4 Dudhia et al. 509

lihood of third molar eruption.7,10-15 Numerous re-searchers have measured third molar inclination and spaceon panoramic radiographs.16-26 Panoramic radiographsare routinely used for assessment of the third molar regionin dental practices. It is important that dental practitionersare mindful of the projection geometry of panoramicradiographs and the distortion and errors inherent in thisimaging modality so as to obtain optimal, accurate diag-nostic information. Computed tomography (CT) has longbeen accepted as a highly accurate method for assessmentof anatomy in all 3 planes, albeit at the expense of asignificantly higher radiation dose compared with pan-oramic radiography. Multidetector CT and reformattedCT images permit accurate assessment of mandibularthird molar angulation in the corrected sagittal and coronalplanes, in addition to providing detailed and accurateinformation regarding the precise location of the inferiordental canal. More recently, cone beam volumetric to-mography (CBVT) has proven to offer dimensionallyaccurate 3-dimensional information at a lower radiationdose than conventional CT.

The purpose of this study was to assess the accuracyof angular measurements made on panoramic radio-graphs by comparison with corresponding angular mea-surements made on corrected sagittal CT images. Theinfluence of bucco-lingual inclination on the observedsagittal inclination will also be determined. Thus, dis-tortion in the third molar region on panoramic radio-graphs can be assessed.

MATERIAL AND METHODSThe images examined in this study were obtained from

a private radiology practice (Queensland Diagnostic Im-aging, Brisbane Private Hospital, Queensland, Australia)over an 18-month period. Patients were referred for a CTscan of their mandibular third molars because the referringpractitioner suspected a very close relationship betweenthe inferior dental canal and at least one of the mandibularthird molars from a panoramic radiograph. Thus, the CTimages were obtained not because of the need to assess theorientation of the third molar, but because of a suspectedrisk of injury to the inferior dental neurovascular bundle.Scans were collected for the study only if a copy of thepanoramic radiograph could be obtained and the imagewas deemed of suitable quality. Patients missing the ad-jacent mandibular second molar were not included in thestudy.

In total, 88 patient scans with 163 mandibular thirdmolars were collected. Panoramic radiographs and theCT scans were de-identified and duplicated. Many, butnot all of the panoramic radiographs where taken by thesame radiology practice that performed the CT exami-nations using a variety of different panoramic ma-

chines. Cases with suboptimal panoramic radiographs

or panoramic radiographs with positioning errors wereexcluded from this study.

Computed tomography imagingComputed tomography scans were obtained using a

multidetector CT Unit (Toshiba Aquilion, Toshiba Cor-poration, Tokyo, Japan) with 0.5-mm axial slices (pitch of1.0 with 0.5-mm reconstruction interval). The scan vol-ume began inferior to the mandibular body, extendingsuperiorly to the middle of the ascending rami. Exposureparameters were set to 135 kV and 60 mA and a low-frequency cutoff was used in the reconstruction algorithm.

All scans were acquired in the axial plane with nogantry tilt. Reformatted images were produced by theCT technicians in the corrected sagittal and coronalplanes for optimal assessment of third molar positionand relationship to the adjacent inferior dental canal (asper a previous study27).

Digitization of imagesPanoramic films and reformatted CT images were

digitized using an Epson Perfection V700 scanner withtransparency adapter (Seiko Epson Corporation, Na-gano, Japan). All images were scanned at 400 dpi ingrayscale, and oriented the same way on the scanner toensure consistency. Individual brightness and contrastadjustments were made before scanning to ensure op-timal visibility of the third molar region. All imageswere saved as jpeg files at the highest-quality setting.

Evaluation of digitized radiographic imagesDigitized images were imported into an x-ray viewing

program (OsiriX; http://www.osirix-viewer.com; Geneva,Switzerland) on an Apple Macintosh personal computer(Apple, Cupertino, CA). Following the correction of im-age orientation, numerous parameters were recorded byone of the authors.

Apparent tilting of the mandibular third molars onthe panoramic radiograph was assessed. The third mo-lar was recorded as being tilted across the arch if therewas separation of the buccal and lingual cusp tips onthe panoramic radiograph.

The angle of inclination between the long axes of thesecond and third molars was recorded from the pan-oramic radiographs and the corrected sagittal reformat-ted CT images. A line bisecting the maximal mesio-distal width of the crown and a line joining the mesialand distal cemento-enamel junction defined the longaxes of the molars (Fig. 1, a). The angle formed be-tween the long axes of the second and third molars wasrecorded using an angle-measuring tool in OsiriX (Fig.1, b). The mesio-distal inclination was recorded asfollows: if the third molar was tilted mesially with

respect to the second molar, the angle was recorded as

OOOOE510 Dudhia et al. April 2011

positive, and if the third molar was tilted distally theangle was negative. On the cross-sectional CT images

Fig. 1. a, A line bisecting the maximal mesio-distal width of thecrown and the line joining the mesial and distal CEJ definesthe long axis of the tooth. b, The measured angle formed by theintersection of the long axes of the second and third molars. c,The long axis of the third molar on the cross-sectional image andthe line defining the buccal cortex. B, buccal; Li, lingual.

the long axis of the third molar was similarly deter-

mined. A line tangent to the buccal cortex was drawnand the angle formed between the long axis of the thirdmolar and the buccal cortex was recorded (Fig. 1, c).Inclination of the third molar toward the buccal cortex wasrecorded as positive and inclination away from the buccalcortex was negative. All measurements were made by thesame author (R.D.) on 3 separate occasions and the authorwas blinded to the previous measurements.

Based on the average measured angle observed onthe panoramic radiograph and the CT images, the teethwere classified using a modification of Winter’s28 orig-inal classification for mandibular third molars (Fig. 2).

Statistical analysisAll radiographic images were obtained from patients

referred for CT imaging because of a perceived closerelationship between the third molar roots and the associ-ated inferior dental canal. Although it is unlikely that thisrelationship would influence any apparent distortion in thepanoramic images, one must be mindful of the exploratorynature of this study when interpreting the results andextrapolating them to other clinical scenarios.

Linear regression analysis of the average panoramicand CT sagittal angulations (the angle formed betweenthe long axes of the mandibular second and third mo-lars) was performed to confirm a direct relationship.Subsequent analysis using the paired Student t test totest for similarity of the measurements was performed.The measured difference between the panoramic andCT sagittal angulations was correlated with the mea-sured cross-sectional angulation of the third molar(measured on the cross-sectional CT images). Further-more, the observation of tilting of the third molar on thepanoramic radiograph was correlated with the mea-sured cross-sectional angulation.

RESULTSEighty-eight patients with 163 impacted mandibular

third molars referred for CT examination were used inthis study. There were 61 females with 110 third molarsand 27 males with 53 third molars. Most patients pre-sented with bilateral mandibular third molars. Only 13cases were unilateral: 12 female and 1 male. The aver-age age was 32 years with a range of 17 to 62 years.The distribution of ages was roughly similar for malesand females.

From the panoramic radiograph, 88 teeth were ob-served to be tilted (T) across the arch, and 75 did notappear to be tilted (NT). A slightly higher proportion oftilted teeth was found in males (58.5% compared with51.8% in females).

From the panoramic radiographs, 22 teeth weredisto-angular, 47 were vertical, 73 were mesio-angular,

and 21 teeth were horizontal. Classification from the

�70 d

OOOOEVolume 111, Number 4 Dudhia et al. 511

CT recorded similar findings: 22 teeth were disto-an-gular, 47 teeth were vertical, 72 teeth were mesio-angular, and 22 teeth were horizontal (Table I). Further

Fig. 2. Cropped panoramic and CT images showing third mvertical (�10 to �10 degrees); c, mesio-angular (� �10 to

analysis revealed that 4 teeth recorded as being disto-

angular on the panoramic radiographs were actuallyvertical on the CT. Additionally, 4 teeth that werevertical on the panoramic radiograph were disto-angu-

lassification: a, disto-angular (�70 to � �10 degrees); b,egrees); d, horizontal (� �70 degrees).

olar c

lar on the CT. Likewise, 5 teeth recorded as being

regress

esio-an

OOOOE512 Dudhia et al. April 2011

vertical on the panoramic radiograph were mesio-an-gular on CT, and 5 teeth that were mesio-angular on thepanoramic radiographs were vertical on the CT. Threeteeth classified as being mesio-angular on the pan-oramic images were horizontal on the CT, and 2 thatwere horizontal on the panoramic radiographs were infact mesio-angular on the corresponding CT images.

Simple linear regression confirmed the presence of adirect linear relationship between the panoramic angu-lar measurements and the corresponding CT angularmeasurements (Fig. 3). The Pearson correlation coeffi-cient (R) was 0.9858 (R2 � 0.9719), indicating a sig-nificant positive correlation between the 2 measure-ments. The average difference between the panoramicmeasurements and the CT measurements was –1.44degrees (SD � 5.75 degrees), indicating that overall thethird molar appeared less mesially inclined on the pan-oramic radiograph compared with the corrected sagittalCT images. The mean for third molars on the right was�1.39 degrees and the mean for third molars on the leftwas �1.49 degrees, and a 2-sample Student t testreturned a P value of .9118, indicating that this differ-ence was not significant. As such, all third molars weregrouped together for the remaining analyses, as the side

Fig. 3. Scatter plot of panoramic angle and CT angle (with

Table I. Classification of third molars from panoramicDA

Classification (�70 to � �10) (�10

Panoramic radiograph 22CT 22

CT, computed tomography; DA, disto-angular; H, horizontal; MA, m

of the third molar is not significant. The high standard

deviation suggested that there is considerable variationin the difference between the angular measurements onthe 2 imaging modalities.

As the primary objective was to assess similaritybetween the panoramic and CT angular measurementsin the mandibular third molar region, the paired Studentt test was used. A P value of .0017 was observedindicating that there are statistically significant differ-ences between the angular measurements obtained fromthe panoramic radiographs and the reformatted CT im-ages. Further investigation revealed that for 99 casesthe third molar was less mesially oriented on the pan-oramic radiograph, with a mean angle of –5.03 degrees(Fig. 4). The remaining 64 cases showed a more mesialinclination of the third molar on the panoramic radio-graph, with a mean angle of 4.12 degrees (Fig. 5). Thus,the overall magnitude of the difference between theangular measurements is between 4 and 5 degrees, butthe direction of the change is variable. No correlationbetween the direction of change and either the mea-sured panoramic or CT angulation was observed.

The bucco-lingual inclination of the third molar asmeasured on the cross-sectional CT images was plottedagainst the difference between the average panoramic

ion line).

graph and CTMA H

) (� �10 to �70) (� �70) TOTAL

73 21 16372 22 163

gular; V, vertical.

radioV

to �10

4747

and CT angular measurements (Fig. 6), and linear re-

OOOOEVolume 111, Number 4 Dudhia et al. 513

gression analysis was performed. No strong associationwas found, and the Pearson correlation coefficient (R)was 0.1617 (R2 � 0.0262).

Apparent tilting across the arch of the third molar onthe panoramic radiograph was recorded against thedifference between the observed average panoramicand CT angulations. A mean difference of �1.16 de-grees was recorded for the teeth that did not appear tobe tilted, and the P value of 0.0631 suggests that thisdifference is only just not statistically significant. Forteeth that were apparently tilted, a mean difference of�1.68 degrees was observed, and the calculated Pvalue of .0118 suggests that this difference is signifi-cant. The average cross-sectional angulation was cal-culated for teeth that did not appear to be tilted acrossthe arch and those that did. The average cross-sectionalangulation for the 75 teeth that did not appear to betilted was �10.83 degrees, whereas the average cross-

Fig. 4. Cropped panoramic (A) and corrected sagittal CT (B)images showing tooth 38 that is less mesially inclined on thepanoramic image compared with on the CT.

sectional angulation for the 88 teeth that did appear to

be tilted was �17.97 degrees (Fig. 7). A 2-sampleStudent t test produced a P value of .0009, indicatingthat this difference is significant.

DISCUSSIONPanoramic radiography is widely accepted as the

imaging modality of choice for assessment of mandib-ular third molars, owing to the large area of visualizedtissues, excellent patient acceptance, and relatively lowradiation exposure. However, inherent distortions intro-duced owing to the image projection geometry mayaffect subsequent interpretation, and in critical casesinsufficiently accurate information is obtained. The in-creased availability of multidetector CT and cone beamtomography, and the dimensionally accurate nature ofthe reformatted images provides practitioners with thehighest quality diagnostic information, but with in-

Fig. 5. Cropped panoramic (A) and corrected sagittal CT (B)images showing tooth 48 that is more mesially inclined on thepanoramic image compared with on the CT (note the impressionof the 48 being tilted across the arch on the panoramic image).

creased radiation exposure and at a higher financial

OOOOE514 Dudhia et al. April 2011

cost. Thus, panoramic radiography is usually the firstchoice for radiographic assessment of impacted third

Fig. 6. Scatter plot of the difference between the panoramiccross-sectional CT (with regression line).

Fig. 7. a, Thirty-eight not showing signs of tilting on the panappears upright. c, Tilted appearance of 48 on the panoramicimage. B, buccal; Li, lingual.

molars, in spite of its limitations.

Many authors have attempted to perform detailed as-sessments of third molars on panoramic radiographs.16-26

T angles against the bucco-lingual inclination measured on

c image. b, The same tooth on the cross-sectional CT imagee. d, Lingual inclination of the 48 on the cross-sectional CT

and C

oramiimag

Assessment of space in the third molar region, third molar

OOOOEVolume 111, Number 4 Dudhia et al. 515

angulation and position, the anatomic relationship with theinferior dental canal, and the prediction of third molareruption have all been discussed. However, there appearsto be little consensus regarding the relative accuracy ofangular measurements in the mandibular third molar re-gion despite the fact that numerous authors perform suchmeasurements routinely. It is not uncommon for dentalpractitioners to subjectively assess the inclination of thethird molar during radiographic interpretation, so the pur-pose of this study was to assess the relative accuracy ofangular measurements made in the mandibular third molarregion on panoramic radiographs, using the correspondingdimensionally accurate reformatted CT images for com-parison. It is important to note that CT examinations wereperformed because of a suspected close relationship be-tween the third molar roots and the adjacent inferior dentalcanal. CT was used, as cone beam tomography scannerswere not available in local imaging centers, and referringsurgeons favored the high-quality images printed on ra-diographic film produced by CT scanning. Furthermore,scan protocols were optimized to reduce patient radiationexposure, and scan field of view was heavily limited toavoid unnecessary radiation exposure. The apparent influ-ence of observed tilting of the third molar across the archon the panoramic radiograph and the bucco-lingual incli-nation observed on the reformatted CT images were alsoconsidered.

A mean difference of �1.44 degrees (SD � 5.75degrees) was noted between the angular measurementsmade on panoramic radiographs and corrected sagittalCT images. This indicates that on average the pan-oramic radiograph underestimated the mesial inclina-tion of the third molar. The observed P value of .0017indicates that this difference is statistically significant.No significant difference was observed based on theside of the third molar (P �.9118). The relatively largestandard deviation suggests that there is considerablevariability in the direction and magnitude of the dis-crepancy between the angular measurements made onthe 2 modalities. Further examination revealed that 99teeth were less mesially inclined on the panoramicradiograph (mean angle of �5.03 degrees, SD � 3.70degrees), and 64 teeth were more mesially inclined(mean angle of 4.11 degrees, SD � 3.50 degrees).

An overall discrepancy of 4 to 5 degrees betweenangular measurements made on panoramic radiographsand reformatted CT images is similar to the findings ofother authors. Sant’Ana et al.24 observed a 5.37-degreediscrepancy in their study, and Peck et al.22 noted adifference of 3.59 degrees between panoramic radio-graphs and CBVT. Interestingly, these studies noted thatpanoramic radiographs always overestimated the mesialinclination of the mandibular third molar, which is in

contrast to the findings in this study. This may be because

of the sample size in this study being significantly largerthan in these previous studies, or that panoramic imagesfrom many (unidentifiable) machines taken by many op-erators were assessed in this study. Previous studies withsmall samples assessed images from a single panoramicmachine only (usually taken by a single operator). Itseems that the panoramic image may project the thirdmolar with either a more mesial or less mesial inclination,although overall the panoramic radiograph seemed to un-derestimate the true mesial inclination.

Bucco-lingual inclination as observed on cross-sec-tional CT images and observed tilting across the arch ofthe mandibular third molar on the panoramic radio-graph were correlated with the difference between theangular measurements in the third molar regions. Teeththat appeared tilted across the arch on the panoramicradiograph, as described by separation of the buccaland lingual cusp tips on the panoramic image, exhibiteda higher mean difference between the angular measure-ments made on the panoramic image and CT image.The mean difference for tilted teeth was �1.68 degrees,whereas the mean difference for teeth that were nottilted was �1.16 degrees. A P value of .0118 suggeststhat the mean difference for tilted teeth is statisticallysignificant and therefore teeth that appear tilted acrossthe arch on the panoramic radiograph are more likely tobe projected at an aberrant angle. The observation oftilting on the panoramic image was also correlated withthe measured bucco-lingual inclination on the cross-sectional CT images. The mean bucco-lingual inclina-tion of the teeth that appeared tilted on the panoramicradiograph was –17.97 degrees, compared with a meaninclination of –10.83 degrees for teeth that did notappear tilted. Once again, the P value of .0009 indicatesthat this difference is significant, and there is reason-ably good correlation between the tilted appearance onthe panoramic radiograph and the bucco-lingual incli-nation observed on CT images.

Very few previous studies have assessed the effect ofbucco-lingual inclination on the panoramic third molarangulation. The finding that altered bucco-lingual incli-nation has some effect on the observed panoramicangulation is consistent with findings in previous stud-ies.23,26 It is difficult to quantify the effects in a clinicalstudy like this one, but it appears that teeth with severebucco-lingual inclination are projected with a signifi-cantly altered angulation on the panoramic radiograph.Thus, care should be exercised when interpreting themesio-distal inclination of third molars on panoramicradiographs where there are signs of apparent tilting ofthe third molar across the arch.

Given the variable accuracy with which panoramicradiographs project the orientation of mandibular third

molars, practitioners should be aware that radiographic

OOOOE516 Dudhia et al. April 2011

observations may not always correlate precisely withthe clinical findings. The tooth may be more mesially orless mesially inclined than is apparent on the panoramicimage. Furthermore, if the apparent inclination differsmarkedly from the actual orientation of the tooth, as-sessment of the likelihood of eruption may be unreli-able. Assessment of the mesial/distal orientation oflower third molars is more likely to be inaccurate whenthe tooth appears tilted across the arch on the pan-oramic radiograph. Tilting on the panoramic imageseems to indicate bucco-lingual tilting of the tooth butthe direction of the tilt cannot be determined. Theobservation of tilting is also likely to alter the plannedsurgical technique. It seems that angular classificationsbased solely on panoramic radiographs do not accu-rately reflect the true classification in some instances.

Although it is neither practical nor advisable to commitall patients with unerupted mandibular third molars to aCT scan, care is required when interpreting the findings inthe third molar region on panoramic radiographs. Pan-oramic radiographs are still useful for preliminary assess-ment and determination of the need for further imaging.With the increasing availability of CBVT and the contin-ual reductions in radiation exposure with successive gen-erations of CBVT scanners and CT scanners, it may soonbe practicable to routinely request accurate 3-dimensionalinformation about impacted teeth to better facilitate diag-nosis and treatment planning.

Special thanks are extended to the CT staff at QueenslandDiagnostic Imaging for CT scanning and image duplication.Measurements were made using OsiriX (http://www.osirix-viewer.com). All statistical analyses were carriedout using StatCrunch (http://www.statcrunch.com; PearsonEducation, Upper Saddle River, NJ, USA) and Minitab(Minitab Incorporated, State College, PA, USA) (v. 15).

REFERENCES1. Badawi-Fayad J, Levy JC, Yazbeck C, Cavezian R, Cabanis EA.

Eruption of third molars: relationship to inclination of adjacentmolars. Am J Orthod Dentofac Orthop 2004;125:200-2.

2. Dachi SF, Howell FV. A survey of 3,874 routine full-mouthradiographs. II. A study of impacted teeth. Oral Surg Oral MedOral Pathol 1961;14:1165-9.

3. Grover PS, Lorton L. The incidence of unerupted permanentteeth and related clinical cases. Oral Surg Oral Med Oral Pathol1985;59:420-5.

4. Morris CR, Jerman AC. Panoramic radiographic survey: a studyof embedded third molars. J Oral Surg 1971;29:122-5.

5. NIH. Consensus development conference for removal of thirdmolars. J Oral Surg 1980;38:235-6.

6. Al-Khateeb TH, Bataineh AB. Pathology associated with im-pacted mandibular third molars in a group of Jordanians. J OralMaxillofac Surg 2006;64:1598-602.

7. Niedzielska IA, Drugacz J, Kus N, Kreska J. Panoramic radiographicpredictors of mandibular third molar eruption. Oral Surg Oral Med OralPathol Oral Radiol Endod 2006;102:154-8.; Discussion:159.

8. Andreasen J, Petersen J, Laskin D. Textbook and color atlas of

tooth impactions. Copenhagen: Munksgaard; 1997.

9. Renton T, Smeeton N, McGurk M. Factors predictive of difficulty ofmandibular third molar surgery. Br Dent J 2001;190:607-10.

10. Ricketts RM. Studies leading to the practice of abortion of lowerthird molars. Dent Clin North Am 1979;23:393-411.

11. Haavikko K, Altonen M, Mattila K. Predicting angulational devel-opment and eruption of the lower third molar. Angle Orthod1978;48:39-48.

12. Shiller WR. Positional changes in mesio-angular impacted mandibularthird molars during a year. J Am Dent Assoc 1979;99:460-4.

13. Venta I, Murtomaa H, Ylipaavalniemi P. A device to predictlower third molar eruption. Oral Surg Oral Med Oral Pathol OralRadiol Endod 1997;84:598-603.

14. Hattab FN, Alhaija ES. Radiographic evaluation of mandibularthird molar eruption space. Oral Surg Oral Med Oral Pathol OralRadiol Endod 1999;88:285-91.

15. Sandhu S, Kaur T. Radiographic evaluation of the status of thirdmolars in the Asian-Indian students. J Oral Maxillofac Surg2005;63:640-5.

16. Benediktsdottir IS, Wenzel A, Petersen JK, Hintze H. Mandib-ular third molar removal: risk indicators for extended operationtime, postoperative pain, and complications. Oral Surg Oral MedOral Pathol Oral Radiol Endod 2004;97:438-46.

17. Bohm B, Hirschfelder U. Localization of lower right molars in apanoramic radiograph, lateral cephalogram and dental CT. JOrofac Orthop 2000;61:237-45.

18. Catic A, Celebic A, Valentic-Peruzovic M, Catovic A, JerolimovV, Muretic I. Evaluation of the precision of dimensional mea-surements of the mandible on panoramic radiographs. Oral SurgOral Med Oral Pathol Oral Radiol Endod 1998;86:242-8.

19. Freisfeld M, Drescher D, Kobe D, Schuller H. Assessment of the spacefor the lower wisdom teeth. Panoramic radiography in comparison withcomputed tomography. J Orofac Orthop 1998;59:17-28.

20. Ganss C, Hochban W, Kielbassa AM, Umstadt HE. Prognosis ofthird molar eruption. Oral Surg Oral Med Oral Pathol 1993;76:688-93.

21. Olive R, Basford K. Reliability and validity of lower third molarspace-assessment techniques. Am J Orthod 1981;79:45-53.

22. Peck JL, Sameshima GT, Miller A, Worth P, Hatcher DC.Mesiodistal root angulation using panoramic and cone beam CT.Angle Orthod 2007;77:206-13.

23. Samawi SS, Burke PH. Angular distortion in the orthopantomo-gram. Br J Orthod 1984;11:100-7.

24. Sant’Ana LF, Giglio FP, Ferreira O Jr., Sant’Ana E, CapelozzaAL. Clinical evaluation of the effects of radiographic distortionon the position and classification of mandibular third molars.Dentomaxillofac Radiol 2005;34:96-101.

25. Wenzel A, Aagaard E, Sindet-Pedersen S. Evaluation of a newradiographic technique: diagnostic accuracy for mandibular thirdmolars. Dentomaxillofac Radiol 1998;27:255-63.

26. Yeo DK, Freer TJ, Brockhurst PJ. Distortions in panoramicradiographs. Aust Orthod J 2002;18:92-8.

27. Mahasantipiya PM, Savage NW, Monsour PA, Wilson RJ. Nar-rowing of the inferior dental canal in relation to the lower thirdmolars. Dentomaxillofac Radiol 2005;34:154-63.

28. Winter G. Principles of exodontia as applied to the impacted mandib-ular third molar. St Louis: American Medical Book Company; 1926.

Reprint requests:

Paul A. Monsour, MdSc, PhDFICD Imaging ClinicSchool of DentistryUniversity of Queensland200 Turbot StreetBrisbane, QLD 4000 Australia

p.monsour@uq.edu.au