Angle Orthodontist, Vol 74, No 2, 2004155

Original Article

The Effects of Differences in Landmark Identification on theCephalometric Measurements in Traditional Versus

Digitized CephalometryYi-Jane Chen, DDS, MSDa; Ssu-Kuang Chen, DDS, PhDb; Jane Chung-Chen Yao, DDS, PhDc;

Hsin-Fu Chang, DDS, MSDd

Abstract: The aim of this study was to explore the effects of differences in landmark identification onthe values of cephalometric measurements on digitized cephalograms in comparison with those obtained fromoriginal radiographs. Ten cephalometric radiographs were randomly selected from orthodontic patients re-cords. Seven orthodontic residents identified 19 cephalometric landmarks on the original radiographs anddigitized images. Twenty-seven cephalometric measurements were computed with a customized computer-aided program. To assess the concordance between cephalometric measurements derived from landmarksidentified on the original radiographs and those from digitized counterparts, the values of 27 cephalometricmeasurements were compared to quantify the absolute value of measurement difference and the interobservererrors between these two methods. We found that the differences of all cephalometric measurements betweenoriginal radiographs and their digitized counterparts were statistically significant. The differences in 21 ofthe 27 cephalometric items were less than two units of measurement (mm or degree), which is generallywithin one standard deviation of norm values in conventional cephalometric analysis. Moreover, statisticallysignificant differences of interobserver errors between the two methods were noted only for seven of the 27cephalometric items. In conclusion, the measurement differences between the original cephalograms and thedigitized images are statistically significant but clinically acceptable. The interobserver errors for cephalo-metric measurements on our digitized cephalometric images are generally comparable with those on theoriginal radiographs. The results of our study substantiated the benefits of digital cephalometry in terms ofthe reliability of cephalometric analysis. (Angle Orthod 2004;74:155161.)

Key Words: Computer-aided cephalometric analysis; Digital imaging; Cephalometric measurement

INTRODUCTIONRapid advances in computer science have led to its wide

application in orthodontics. Computer-based filing systemsfor patients records have the benefits of image storage,

a Lecturer, Department of Orthodontics, School of Dentistry, Col-lege of Medicine, National Taiwan University, Taipei, Taiwan, ROC.

b Associate Professor, Division of Oral Radiology, School of Den-tistry, College of Medicine, National Taiwan University, Taipei, Tai-wan, ROC.

c Assistant Professor, Department of Orthodontics, School of Den-tistry, College of Medicine, National Taiwan University, Taipei, Tai-wan, ROC.

d Associate Professor, Department of Orthodontics, School of Den-tistry, College of Medicine, National Taiwan University, Taipei, Tai-wan, ROC.

Corresponding author: Ssu-Kuang Chen, DDS, PhD, School ofDentistry, College of Medicine, National Taiwan University, No. 1,Chang-Te Street, Taipei 100, Taiwan, ROC(e-mail: cskchen@ntu.edu.tw).Accepted: May 2003. Submitted: May 2003.q 2004 by The EH Angle Education and Research Foundation, Inc.

transmission, and processing.1 Many commercially avail-able programs have been developed to perform cephalo-metric analysis directly on the screen-displayed digital im-age.2,3 Computer-aided cephalometric analysis on digitizedcephalogram could substantially reduce the potential errorsin the use of digitizing pads and eliminate the productionof hardcopies of digitally born images for conventionalcephalometric analysis. Actually, a radiographic system fordirectly taking digitized cephalograms at a reduced radia-tion dose is available.4,5 Otherwise, the conventional ceph-alometric radiographic films could be transformed into adigital format by a film scanner.

The errors in cephalometric analysis are composed ofsystematic errors and random errors; the latter involvestracing, landmark identification, and measurements. Com-puter-aided cephalometric analysis can totally eliminate themechanical errors in drawing lines between landmarks andin measurements with a protractor. Earlier studies revealthat computer-aided cephalometric analysis does not intro-duce more measurement error than hand tracing, as long as

156 CHEN, CHEN, YAO, CHANG

Angle Orthodontist, Vol 74, No 2, 2004

TABLE 1. The 27 Cephalometric Measurements Used in This StudySNASNBANBA-NvPog-NvNAPogWitsabSN-FHSN-OP

Angle determined by points S, N, and AAngle determined by points S, N, and BAngle determined by points A, N and BDistance from point A to Nv line (perpendicular to FH plane from point N)Distance from point Pog to Nv line (perpendicular to FH plane from point N)Angle determined by points N, A, and PogDistance between the projections of point A and B onto occlusal planeDistance between the projections of point A and B onto FH planeAngle determined by SN plane and Frankfort horizontal (FH) planeAngle determined by SN plane and occlusal plane

SN-MPUFH/LFHAr-AAr-GnA-GnAr-A/Ar-GnAArGnAGnArArAGnUI-SN

Angle determined by SN plane and mandibular planeRatio of upper facial height (N-Ans) to lower facial height (Ans-Me)Distance between point Ar and ADistance between point Ar and GnDistance between point A and GnRatio of Ar-A to Ar-GnAngle formed by points A, Ar, and GnAngle formed by points A, Gn, and ArAngle formed by points Ar, A, and GnAngle formed between axis of upper incisor to SN plane

UI-NPogUI-LILI-OPLI-MPaublul

Distance between upper incisor edge to lower facial plane (N-Pog)Angle formed by the intersection of tooth axis of upper incisor and lower incisorAngle formed by the intersection of lower incisor axis to occlusal planeAngle formed by the intersection of lower incisor axis to mandibular planeDistance between the projections of point A and upper incisal edge onto FH planeDistance between the projections of point B and lower incisal edge onto FH planeDistance between the projections of upper and lower incisal edges onto FH plane

landmarks are identified manually.6,7 However, the incon-sistency in landmark identification is still an importantsource of random error in computer-aided digital cephalom-etry. For digital cephalometry to be a better tool in clinicalorthodontics, the cephalometric analysis, represented bywidely used linear and angular measurements, must be ascomparable and reliable as it is on conventional radiograph-ic film. In the previous study, it was demonstrated that thedifferences in landmark identification between originalcephalometric radiographs and their digitized counterpartswere statistically significant.8 This difference may be themajor source of differences in the determination of linearas well as angular measurements on original radiographsand their digitized counterparts. The purpose of this studyis to explore the effect of the differences in landmark iden-tification on the values of cephalometric measurements ondigitized cephalograms in comparison with those obtainedfrom original radiographs.

MATERIALS AND METHODSThe criteria to randomly select cephalometric radio-

graphs and the way to perform cephalometric analysis havebeen described in detail in the previous study.8 In brief, 10conventional cephalometric radiographs were selected fromorthodontic patients records, scanned into the digital for-mat with a roller-type scanner (VXR-12, Vidar SystemCorp. Herndon, Va), and displayed on a high-resolutionmonochromic monitor (Image Systems, Hopkins, Minn).The digitized image resolution was 150 dpi with 256 gray

levels. The 19 landmarks were identified on original radio-graphs by manual dot tracing and then directly identifiedon their monitor-displayed digitized counterparts by sevenorthodontic residents with an intervening time interval ofmore than one week. The landmark identification on orig-inal radiographs was performed on a superimposed trans-parent film, which was scanned and processed in the sameway as that on digitized cephalometry. The 27 cephalo-metric measurements listed in Table 1, including 13 linearand 14 angular measurements, were calculated from the co-ordinates of 19 cephalometric landmarks by computer-aidedcephalometric program. Most of the items are the popularand widely accepted measurements, except the items in-cluded in vector analysis9 and jaw triangle analysis,10which are routinely used in our department (Figure 1). Foreach cephalometric item, the absolute values of the differ-ences between the values derived from the original radio-graphs and those from their digitized counterparts were as-sessed. The mean and standard deviation of cephalometricmeasurement differences were calculated. Moreover, themean value for each of the aforementioned 27 cephalo-metric measurements derived by the seven observers wasdefined as the gold standard, which was used to deter-mine the interobserver errors of measurements on originalfilm and digitized image. The interobserver error was usedas an indicator to determine the reliability of each cepha-lometric measurement. Consequently, the reliability ofcephalometric measurement could be quantified and com-pared between these two methods.

157CEPHALOMETRIC MEASUREMENTS IN TRADITIONAL VS DIGITIZED CEPHALOMETRY

Angle Orthodontist, Vol 74, No 2, 2004

FIGURE 1. The printout record of our computer-aided cephalometric analysis.

TABLE 2. Differences of Skeletal Cephalometric MeasurementsBetween the Two Methods and Statistical Analysis for InterobserverErrorsa

ItemMeasurement

Differenceb

InterobserverError on Original

Radiograph

InterobserverError on Digital

ImageSNASNBANBA-NvPog-NvNAPogWitsabSN-FHSN-OP

1.43 6 1.641.07 6 1.360.95 6 0.881.92 6 1.993.09 6 2.811.75 6 1.762.05 6 3.551.28 6 1.191.61 6 1.521.82 6 1.84

1.13 6 1.290.95 6 1.040.62 6 0.560.99 6 0.831.51 6 1.501.02 6 1.051.64 6 2.510.95 6 0.841.05 6 1.051.29 6 1.01

0.90 6 1.110.65 6 0.74c0.46 6 0.491.12 6 1.161.58 6 1.061.13 6 1.120.95 6 0.85c0.83 6 0.520.85 6 0.711.10 6 1.08

SN-MPUFH/LFHAr-AAr-GnA-GnAr-A/Ar-GnArAGnAArGnAGnAr

1.56 6 1.441.16 6 1.301.29 6 2.921.13 6 2.831.52 6 2.831.04 6 1.041.39 6 1.330.88 6 0.831.02 6 0.95

1.17 6 0.990.57 6 0.620.61 6 0.550.70 6 0.640.54 6 0.540.54 6 0.550.72 6 0.610.48 6 0.390.55 6 0.56

1.09 6 0.960.85 6 0.891.06 6 2.441.08 6 2.381.44 6 2.05c0.63 6 0.761.01 6 1.010.70 6 0.60c0.63 6 0.68

a Linear measurement in millimeters, angular measurements indegrees.

b The differences for all items reached significant level (P , .05).c Significant difference of interobserver errors between the two

methods was detected by paired t-test (P , .05).

STATISTICAL ANALYSISTo analyze the variation of cephalometric measurement

differences, the mean and standard deviation of the ceph-alometric measurement differences between the originalfilms and the digitized images were calculated for each ofthe 27 items. The statistical significance of measurementdifferences and the comparison of interobserver errors be-tween the two methods were checked by paired Studentst-test.

RESULTSThe mean difference and standard deviation for each of

the 19 skeletal measurements on original radiographs andtheir digitized counterparts are presented in Table 2. Themean differences range from 0.888 to 1.828 for angularmeasurements and from 1.13 to 3.09 mm for linear mea-surements. The comparable results of eight dental measure-ments are shown in Table 3. The mean differences rangedfrom 4.088 to 7.688 for angular measurements and from0.84 to 1.27 mm for linear measurements. To our surprise,the angular dental measurements with greater values ofmeasurement differences between the two methods wereexclusively the items indicating the tooth axis of the incisor.However, the measurement differences were less in the lin-ear dental measurements, which were delineated from anincisal edge to a specific reference plane and indicated theprotrusive position of the incisor (eg, UI-N Pog, au, andbl). All the measurement differences for each of these 27cephalometric items were statistically significant (P , .05).However, the values of 21 in the 27 items were less thantwo units of measurement (mm or degree), which is gen-

erally within one standard deviation of norm values in con-ventional cephalometric analysis.810 To have a better over-view, the distribution of cephalometric items can begrouped according to the level of measurement differences

158 CHEN, CHEN, YAO, CHANG

Angle Orthodontist, Vol 74, No 2, 2004

TABLE 3. Differences of Dental Cephalometric Measurement Be-tween the Two Methods and Statistical Analysis for InterobserverErrorsa

ItemMeasurement

Differenceb

InterobserverError on Original

Radiograph

InterobserverError on Digital

ImageUI-SNUI-LILI-OPLI-MPUI-NPogaublul

4.89 6 6.867.68 6 4.744.08 6 6.004.16 6 5.730.84 6 1.221.27 6 1.390.90 6 0.991.08 6 1.93

3.16 6 3.894.26 6 4.232.41 6 3.111.79 6 1.370.49 6 0.630.91 6 0.930.67 6 0.510.57 6 0.77

3.41 6 4.914.36 6 5.113.43 6 4.223.70 6 4.33c0.69 6 0.77c0.94 6 1.020.80 6 0.881.00 6 1.36c

a Linear measurement in millimeters, angular measurements indegrees.

b The differences for all items reached significant level (P , .05).c Significant difference of interobserver errors between the two

methods was detected by paired t-test (P , .05).

TABLE 4. The Distribution of Cephalometric Items According tothe Level of Measurement Difference (md) Between Traditional andDigitized Cephalogramsa

md , 1.0 1.0 , md , 1.5 1.5 , md , 2.0 2.0 , md

ANBUI-NPogAArGnbl

SNASNBUFH/LFHAr-AAr-GnAr-A/Ar-GnArAGnAGnArauulab

A-NvNAPogSN-FHSN-OPSN-MPWitsAGn

Pog-NvUI-SNUI-LILI-OPLI-MP

a Linear measurement in millimeters, angular measurements indegrees.

between traditional and digitized cephalograms (Table 4).The four items, ANB, UI-NPog, AArGn, and bl, had mea-surement differences of less than one unit (mm in linearmeasurement and degree in angular measurement). Theitems with measurement differences of more than two unitswere Pog-Nv, UI-SN, UI-LI, LI-OP, and LI-MP, most ofwhich were measurements indicating the tooth axis incli-nation. The linear dental measurements to evaluate incisalposition by the distance between the incisal edge and aspecific reference plane were shown to have measurementdifferences in the middle level (between one and two units).The differences of angular measurements in jaw triangleanalysis (AArGn, ArAGn, and AGnAr) were generallysmaller than the measurements that indicate the facial di-vergence by the intersection of two reference planes, whichwere constructed by four landmark points (SN-FH, SN-OP,and SN-MP).

With regard to the comparison of interobserver errorsbetween these two methods, seven of the 27 items were

significantly different (Tables 2 and 3). Generally, the meanvalue and standard deviation of interobserver error of ceph-alometric measurements in the digitized images was com-parable with those in the original radiographs. The meaninterobserver error in the original radiographs ranged from0.488 to 4.268 for angular measurements and from 0.49 to1.64 mm for linear measurements. The mean interobservererror in the digitized images ranged from 0.468 to 4.368 forangular measurements and from 0.69 to 1.58 mm for linearmeasurements. The interobserver errors in the original ra-diographs and digitized cephalogram was generally com-parable with or slightly less than the measurement differ-ences between the two methods. It was noted that the in-terobserver errors on original radiographs and digitizedcephalogram were less than one mm or degree in 16 of the27 cephalometric measurements. All the three angular skel-etal measurements (SN-FH, SN-OP, and SN-MP) that in-dicate the facial divergence by two reference planes dem-onstrated interobserver errors of more than 18. In both mo-dalities, the measurements with larger interobserver error(more than two units) were found to be the measurementsrepresenting the tooth axis inclination (UI-SN, UI-LI, LI-OP, and LI-MP). The measurement with the least interob-server error, eg, the most reliable measurement, was AArGnin original radiographs and ANB in digitized counterparts.In the 19 skeletal measurements, the statistically significantdifferences of interobserver errors between original radio-graphs and digitized images were demonstrated in two an-gular measurements (SNB and AArGn) and two linear mea-surements (Wits and ab). Surprisingly, the interobservererrors of SNB and Wits in digitized cephalometry weresmaller than those derived from original radiograph. Withregard to dental measurements, three items (LI-MP, UI-NPog, and ul) of all the eight items demonstrated significantinterobserver errors between these two modalities (Table 3).

DISCUSSION

The major errors associated with conventional cephalom-etry include projection errors and tracing errors. The mostimportant source of tracing errors is uncertainty in land-mark identification, and intraobserver error is generally lessthan interobserver error.6,11 The mechanical errors intro-duced by drawing lines between landmarks manually andby measuring with a ruler and protractor were common inconventional cephalometric analysis. However, these errorswere eliminated from our study because the analysis wasperformed by computer-aided cephalometric program.When we take advantage of digital cephalometry, it is im-portant to be certain that the digitized image yields the sim-ilar performance to conventional radiographic film in termsof cephalometric measurements. The major source of errorin the determination of linear and angular cephalometricmeasurements in this study was supposed to be the errorsin landmark identification, which were assessed in the pre-

159CEPHALOMETRIC MEASUREMENTS IN TRADITIONAL VS DIGITIZED CEPHALOMETRY

Angle Orthodontist, Vol 74, No 2, 2004

vious work.8 In our previous study, the overall differencesof landmark identification between the two methods, con-ventional radiographs and digitized images, were statisti-cally significant. The extent of the difference for each land-mark depended on the radiographic complexities, whichwere also associated with the reliability of the landmarks.The representation of head films and observers should beconsidered as possible sources of error when comparingcomputer-aided cephalometric analysis based on conven-tional radiographs and digitized images. The head filmsused in this study were randomly selected from the patientsfiles. They were representative of the films that we consid-ered satisfactory for routine clinical use. The observers inthis study were the residents under the three-year postgrad-uate training program and were considered to be competentclinicians with average training in cephalometrics.

In clinical orthodontics, cephalometric analysis has longbeen used as an important clinical tool in diagnosis, treat-ment planning, and evaluation of growth or treatment re-sults. Many parameters were proposed to analyze the rela-tionship of teeth to teeth, teeth to jaws, jaws to cranial base,and the interjaw relationship. For example, the anteriopos-terior jaw relationship could be evaluated by many linearand angular measurements in cephalometrics. Linear mea-surements may be affected by the inclination of the refer-ence line, and angular measurements cannot indicate cor-rectly the jaw relationship in the case of extreme facialdivergency.12 Therefore, it is reasonable to evaluate a set ofstructural relationships by multiple cephalometric parame-ters rather than by a single parameter. This is the reasonwhy we included as many as 27 measurements in our cus-tomized cephalometric analysis to evaluate the routine or-thodontic cases. However, the dependence of measurementerror should be considered during the interpretation of mul-tiple cephalometric measurements. When a landmark com-mon to a pair of measurements is reused in computing eachof those measurements, the significant correlation betweenmeasurements may result from the error of measurement,which is contributed to the value for measurements. Hous-ton et al13 had discussed that the error of identifying a com-mon landmark between linear or angular variables may re-sult in a purely topographic correlation between them,which may exaggerate or attenuate a true biologic corre-lation. The most practical way to avoid such error is tomeasure the two variables independently on a separate trac-ing of each structure.

In terms of random errors of cephalometric measure-ments, the effect of projection errors caused by incorrectpatient positioning should be considered. The effect of in-correct patient positioning on linear and angular measure-ments was reported by Achlqvist et al.14,15 When the mis-alignment of the patients head was less than 658, the errorswere generally less than 1% in length measurement and lessthan 618 in angle distortion. A misalignment of the pa-tients head of more than 58 is easily detected and should

be corrected immediately. This information means that thepercent error of cephalometric measurements from error inlandmark identification might be greater than that from pro-jection error. Many previous studies have shown that it isimpossible to estimate the positions of landmarks withouterror. However, we should make efforts to minimize theeffect of error in landmark identification on the cephalo-metric measurements, especially the items with inheritedlower reliability in digitized cephalometry. The impact ofthe error in landmark identification on the cephalometricmeasurements can be considered in different aspects. Thefirst is the magnitude of the error inherited in the identifi-cation procedure for the specific landmark, which was in-vestigated in the previous work. The average value of mea-surements by all observers was used as the gold standardfor a specific landmark to quantify the actual error. Thesecond is the distance between two landmarks, which areconnected to construct a reference plane or calculate a givenlinear measurement. In the case of linear measurement, theshorter the line segment measured, the greater the percent-age of error produced by a given measurement error. Thistopographic problem is also true for angular measurements.The closer the points in the construction of an angle, thegreater will be the impact of measurement error on the an-gular measurement calculated.

In this study, all the cephalometric measurements dem-onstrated significant differences between conventional ra-diographs and digitized cephalograms. The differences indental measurements were generally larger than those in theskeletal measurements, especially the angular dental mea-surements. In the previous study, the significant differencesof interobserver error in landmark locations between thesetwo methods were demonstrated only at Po, Ar, ANS, andUM. However, these four landmarks were involved in thedefinition of 19 items of the 27 cephalometric measure-ments in our study. In this study, the significant differencesof interobserver error were noted only in seven of the 27cephalometric measurements. It means that the high reli-ability of landmark identification in digitized cephalogramimplies the comparable reliability of the subsequently de-rived cephalometric measurements. In our study, each ofthe 27 measurements calculated from a total of 19 land-marks involved a minimum of three landmarks for a givenmeasurement. In other words, each landmark estimationwas used in more than four measurements and contributedits inherent estimating error to the computation of eachmeasurement. In this study, the statistically significant re-liability of SNB and Wits in digitized cephalometry com-pared with original radiographs was noted. This findingmay be partly explained by the complex interaction amongtheir relevant landmarks. Although statistically significantdifferences were detected in all cephalometric measure-ments in the comparison of conventional cephalograms andtheir digitized counterparts, their clinical significance

160 CHEN, CHEN, YAO, CHANG

Angle Orthodontist, Vol 74, No 2, 2004

should be referred to the standard deviation of norm valuesfor each item.16

The items with relatively larger measurement differencesand a wide range of variation were angular measurementsreflecting the axis of the upper and lower incisors (UI-SN,UI-LI, LI-OP, and LI-MP). The differences in these dentalmeasurements may be associated with their larger interob-server errors and wider range of variation in both originaland digitized modalities. The short distances among land-mark points used to construct the tooth axis of upper andlower incisors could explain the larger interobserver errorsof dental angular measurements. In other words, the moreclosely located two landmark points are, the greater theangular measurement error tends to be. This geometric re-lationship was reported and discussed by other researchers.Nagasaka et al17 have illustrated the theoretical relationshipbetween the interlandmark distance and the possible an-gular measurement error when the linear measurement erroris set at a certain level.

In the experimental design of our study, the best estimatefor each measurement or gold standard for determining theinterobserver errors was defined as the mean value of mea-surements from seven different observers. The interobserv-er error was used as a variable to determine the reliability,ie, the dispersion of error around the best estimate for eachmeasurement.The interobserver errors of the 27 cephalo-metric items in original cephalometric radiographs and dig-itized images must be exclusive results of error in landmarkidentification because identical data processing was used inboth methods in the study. Every effort should be made tominimize the errors of measurements in cephalometry.However, the quest for precision should not obscure thevalidity of some cephalometric landmarks and measure-ments.16 More replications of landmark identification mayhelp reduce their random errors and, subsequently, the er-rors in cephalometric measurements.

The source of error in landmark identification associatedwith the complexity of radiographic images is the difficultyof delineating a landmark on a curved anatomical bound-ary. We may expect that the powerful tool of digitized im-age processing could help with landmark identification onpoorly defined structures. However, it was reported that thelandmark reliability in digitized radiographs of lower qual-ity could not be improved by digital processing.18 Instead,we should make efforts to improve the precision of land-mark identification by enhancing the image quality at thebeginning. The image revealed on radiographic film is con-sidered an analog, which means that the gray level is con-tinuous. A digital image is composed of many picture el-ements (pixels), which are represented by discrete pixel val-ues. The quality of a digital image strongly depends on boththe number of pixels and the number of gray levels. Thereliability of landmark identification on digital images witha pixel size of 0.47 mm was demonstrated to be inferior tothat on conventional radiographs.19 However, the reliability

on a digitized cephalogram with a pixel size of 0.03 mmwas better than that on original radiographs with conven-tional equipment. The cephalometric radiographs used inthis study were randomly selected and represented the qual-ity of daily routine work. For the consideration of bothdigital image quality and their memory size, the image res-olution was set at 150 dpi with a pixel size of 0.17 mm.

The reliability of landmark identification is expected tobe affected by the image quality. There are several ways ofacquiring a digitized cephalometric image, and the imagequality would depend on how the image was acquired. Inthis study, the digitized images were secondarily capturedby a film scanner from the images on conventional films.Inevitably, image signal deterioration would occur in thedigitization process of scanning and transformation intodigital format. In this case, the quality of the digitized im-ages would be less than that of the original images on film.However, the result of this study implied that the parametersetting for our digitized cephalogram was almost adequatein terms of performance of cephalometric analysis, whichwas demonstrated by the low level of measurement differ-ences and the generally comparable interobserver errors be-tween original and digitized cephalograms. The rapid tech-nological advances in equipment in digital dental radiologyhave led to the increased popularity of direct digital ceph-alometric images, which are either captured by computedradiography or by digital radiography. How directly ac-quired digital cephalometric measurements vary from tra-ditional ones should be further studied.

CONCLUSIONSIn our computerized digital cephalometric analysis, the

differences between the measurements derived from thelandmarks identified on original cephalometric radiographsand those identified on their digitized counterparts were sta-tistically significant but clinically acceptable. The interob-server errors of cephalometric measurements on digitizedimages were generally comparable with those from originalradiographs. In conclusion, the results of our study sub-stantiated the benefits of digital cephalometry in terms ofthe reliability of cephalometric analysis.

ACKNOWLEDGMENTThis work was supported, in part, by a research grant from the

National Science Council of Taiwan (NSC-90-2212-E-002-205).

REFERENCES1. Forsyth DB, Shaw WC. Digital imaging of cephalometric radi-

ology. Part I: advantages and limitation of digital imaging. AngleOrthod. 1996;66:3742.

2. Baskin HN, Cisneros GJ. A comparison of two computer ceph-alometric programs. J Clin Orthod. 1997;31:231233.

3. Gotfredsen E, Kragskov J, Wenzel A. Development of a systemfor craniofacial analysis from monitor-displayed digital images.Dentomaxillofac Radiol. 1999;28:123126.

161CEPHALOMETRIC MEASUREMENTS IN TRADITIONAL VS DIGITIZED CEPHALOMETRY

Angle Orthodontist, Vol 74, No 2, 2004

4. Naslund EB, Kruger M, Petersson A, Hansen K. Analysis of low-dose digital lateral cephalometric radiographs. DentomaxillofacRadiol. 1998;27:136139.

5. Geelen W, Wenzel A, Gotfredsen E, Kruger M, HanssonL-G. Re-producibility of cephalometric landmarks in conventional film,hardcopy and monitor-displayed images obtained by the storagephosphor technique. Eur J Orthod. 1998;20:331340.

6. Gravely JF, Benzies PM. The clinical significance of tracing errorin cephalometry. Br J Orthod. 1984;11:4448.

7. Enlow DH, Hans MG. Essentials of Facial Growth. Philadelphia,Pa: WB Saunders Company; 1996: 259260.

8. Chen YJ, Chen SK, Chan HF, Chen KC. Comparison of landmarkidentification in traditional versus computer-aided digital cepha-lometry. Angle Orthod. 2000;70:387392.

9. Chen KC, Chen MC, Wu KM, Chan HF, Chen YJ, Yuan ST. Thevector analysis of the anteroposterior relationship among the api-cal bases and the incisiors. Chin Dent J. 1989;8:620.

10. Chen YJ, Hong YL, Wu KM, Chen MC, Chan HF, Chen KC.Jaw triangle analysis: an adjuvant diagnostic tool in orthodontics.Chin Dent J. 1993;12:5670.

11. Cohen AM. Uncertainty in cephalometrics. Br J Orthod. 1984;11:4448.

12. Williams S, Leighton BC, Nielsen JH. Linear evaluation of thedevelopment of sagittal jaw relationship. Am J Orthod. 1985;88:235241.

13. Houston WJB, Maher RE, McElroy D, Sherriff M. Sources oferror in measurements from cephalometric radiographs. Eur J Or-thod. 1986;8:149151.

14. Ahlqvist J, Eliasson S, Welander U. The effect of projection errorson cephalometric length measurements. Eur J Orthod. 1986;8:141148.

15. Ahlqvist J, Eliasson S, Welander U. The effect of projection errorson angular measurements in cephalometry. Eur J Orthod. 1988;10:353361.

16. Baumrind S, Frantz RC, Frantz RC. The reliability of head filmmeasurements. 2. Conventional angular and linear measures. AmJ Orthod. 1971;60:505517.

17. Nagasaka S, Fujimura T, Segoshi K. Development of a non-ra-diographic cephalometric system. Eur J Orthod. 2003;25:7785.

18. Macri V, Wenzel A. Reliability of landmark recording on film anddigital lateral cephalograms. Eur J Orthod. 1993;15:137148.

19. Cohen AM, Linney AD. A low cost system for computer-basedcephalometric analysis. Br J Orthod. 1986;13:105118.