Validity of radiographic measurement of interproximal bone loss

Vol. 85 No. 1 January 1998 iii ~

=i~I

I

=

! =iilI ~:iiililiiI!i:IiI~i Iili,,i ~i!iliI IiI i i ! i I !

i=I~II i!i!!ii=III!!II Iiii =#!1111iiliili IIIilI!!ii !I =iilii=i !II~ = =Iii II ~'II

ORAL AND MAXILLOFACIAL RADIOLOGY

Editor: Sharon L. Brooks

Validity of radiographic measurement of interproximal bone loss Peter Eickholz, Dr Med Dent, a Ti-Sun Kim, Dr Med Dent, b Douglas K. Benn, DDR, PhD, BDS, M Phil, c and Hans J. Staehle, Dr Med, Dr Med Dent, d Heidelberg, Germany, and Gainesville, Fla. RUPRECHT-KARLS-UNIVERSITAT AND UNIVERSITY OF FLORIDA

The aim of the present study was to compare radiographic assessmentof interproximal bone loss using a Ioupe with a 0.1 mm calibrated grid and a computer-assisted analysis system (LMSRT). In 35 patients suffering from untreated advanced periodontal disease, 62 standardized radiographs were taken presurgically. The horizontal and vertical angulation difference of the central beam from the orthoradial projection was calculated for each radiograph. At the time of surgery, for 115 interproximal defects, the distances from the cementoenamel junction (CEJ)to alveolar crest (AC), and CEJ to bottom of the bony defect (BD) were measured. In all radiographs, the linear distances CEJ to AC, and CEJ to BD were assessed using a Ioupe and LMSRT. Comparison between radiographic and intrasurgical assessmentswas performed using paired t-tests. A stepwise multiple linear regression analysis was used to evaluate factors that influence the discrepancy between radiographic and intrasurgical measurements. Both analyzing techniques underestimated interproximal bone loss as compared with intrasurgica[ measurements (CEJ-AC: Ioupe: 0.86 _+ 1.84 mm [p < 0.001]; LMSRT: 0.58 _+ 1.86 mm [p < 0.005]; CEJ-BD: loupe: 1.22 _+2.33 mm [p < 0.001]; LMSRT: 0.80 _+2.09 mm [p < 0.001]). LMSRT underestimated interproximal bone loss significantly less than the Ioupe (p < 0.001). The difference between LMSRT and intrasurgical assessmentswas modulated by the factors of vertical and horizontal angulation difference and defect depth (p < 0.1). Orthoradial projection reduced underestimation of radiographic, assessmentof bone, loss. LMSRT underestimated interproximal bone loss to a lesser extent than conventional evaluation by loupe. (Oral Surg Oral Med Oral Pathol Oral

Radiol Endod 1998;85:99-106)

Along with marginal inflammation, periodontal pocket formation, and attachment loss, alveolar bone loss is a primary feature of periodontitis. The height of the alveolar bone may be evaluated by intrasurgical inspection or, less invasively, by radiographic examination; however, radiographic assessment tends to underestimate the amount of bone loss. 1-6 Changes of mineralized tissue like alveolar bone may be detected radiographically from consecutive radiographs. Projection geometry of serial radiographs should be standardized to minimize measurement errors. 7-9 Such errors are often difficult to distinguish from biologic changes. Advanced radiographic analysis by measurement of linear disaLecturer, Department of Operative Dentistry and Periodontology, Dental School. bScientific Assistant, Department of Operative Dentistry and Periodontology, Dental School. CAssociate Professor, Department of Oral Surgery and Diagnostic Sciences, Dental College. dprofessor, Department of Operative Dentistry and Periodontology, Dental School. Received for publication Mar. 6, 1997; returned for revision May 1, 1997; accepted for publication June 16, 1997. Copyright © 1998 by Mosby, Inc. 1079-2104/98/$5.00 + 0 7/16/84769

tances 1°,11 or subtraction 12,13 require highly standardized projection. Whereas subtraction analysis can be used only to assess differences between at least two radiographs, linear measurements can be used to assess bone status on single radiographs. However, even in single radiographs, projection geometry may influence validity of interpretation. The aims of the present study were to (1) assess the validity of linear measurements of interproximal bone loss on radiographs using a loupe with 0.1 mm graduations and using a computer-assisted analyzing tool (LMSRT), (2) compare the validity of both measurement methods, and (3) identify the factors influencing the validity of the superior method.

MATERIAL AND METHODS Patients. Thirty-five patients (16 men and 19 women) suffering from moderate to advanced untreated periodontal disease that were scheduled for periodontal treatment in the Section of Periodontology, Department of Operative Dentistry and Periodontology, Dental School, University of Heidelberg took part in the present study. 99

100 Eickholz et aL

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY

January 1998

Fig. 1. Length of the radiographic image of the maxillary wire on the filmholder IR; distal site of the second right maxillary premolar: cementoenamel junction (CEJ), alveolar crest (AC) and most apical extension of the bony defect (BD).

They ranged from 34 to 64 years in age. Structure, risk s , and benefits of the diagnostic and therapeutic procedures were explained to all patients and consent was obtained.

Radiographic examination. After completion of initial periodontal treatment including oral hygiene instruction and scaling, standardized bitewing radiographs were taken of teeth exhibiting vertical (intrabony defects) or horizontal (degree l]/III furcation lesions) interproximal bone loss. Modified film holders (VIP 2 Film Positioning, UpRad Corp., Fort Lauderdale, Fla.) were used. 6,7,14 The design of these filmholders and the assessment of angulation differences between consecutive radiographs obtained with them have been described extensively elsewhere6,VA4; therefore, only a brief description is provided. Two orthodontic wires were placed on the mandibular side of the filmholder at a specified position. Shadows of these wires were cast on to the radiographs (Figs. 1 and 2). From the distances between the

Fig. 2. Vertical (av) and horizontal (ah) distance of the images of the mandibular wires on the filmholder; mesial site of the second right mandibular premolar. Instead of the cementoenamel junction (CEJ) that was extinguished by restorative treatment, the margin of the restoration was taken as landmark; alveolar crest (AC) and most apical extension of the bony defect (BD).

images of these wires on a radiograph, the vertical and horizontal angulation difference between the central beam and the orthoradial projection could be calculated. Intraoral dental films (Ultraspeed 31 x 41 ram, Eastman Kodak Co, Rochester, N.Y.) size 2 were exposed to an x-ray source (Heliodent 70, 70 kV, 7 mA, Siemens, Benheim, Germany) and developed under standardized conditions (Periomat, Dtirr Dental GmbH, Bietigheim-Bissingen, Germany).

Clinical examinations. Immediately before surgery, the Gingival Index (GI) and Plaque Index (PII) 15 were assessed at six sites per tooth. Probing pocket depths (PD) and vertical probing attachment levels (PAL-V) were measured to the nearest 0.5 m m using a straight periodontal probe (PCPUNC 15, Hu Friedy, Chicago, Ill.). After reflection

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY Volume 85, Number 1

Eickholz et al. 101

Table I. Clinical parameters of 115 interproximal lesions and vertical as well as horizontal angulation difference of the central beam from the orthoradial projection of 62 radiographs Clinical parameters

M e a n _+ S D Range

Angulation difference~degree

G1

PII

PD (ram)

PAL (mm)

Vertical

Horizontal

1.58 _+ 0 . 6 0

0.37 -+ 0.53

5 . 4 6 ___ 1.94

6.01 _+ 2.31

2.03 + 1.56

0.78 + 0 . 7 0

0.0 - 3.0

0.0 - 2.0

2 . 0 - 11.5

1.5 - 13.5

0.0 - 6.0

0.0 - 2.9

Mean + standard deviation (SD); range (minimum - maximum). Gingiva index, G1; plaque index, Pll; probing depth, PD; probing attachment level, PAL.

Table II. Intrasurgical measurements and radiographic parameters (evaluation by loupe and computer-assisted [LMSRT]) of 115 interproximal lesions Intrasurgical measurements (mm)

Radiographic parameters (mm)

Distances

Mean + SD

Range

Loupe mean +_SD

Range

LMSRT mean + SD

Range

CEJ-AC

4 . 3 2 _+ 1.73

1.5 - 12.0

3.46 + 1.70"

0.3 - 8.2

3.79 + 1.85 t

0.5 - 10.0

CEJ-BD

7.31 _+ 2 . 8 2

2.5 - 18.0

6.09 _+ 2.28*

2.4 - 16.2

6.52 + 2 . 4 8 *

2.4 - 15.2

Mean _+ standard deviation (SD); range (minimum - maximum). Cementoenamel junction, CEJ; alveolar crest, AC; apical extension of the bony defect, BD. *Statistically significantly different from intrasurgical measurements (p < 0.05). *Statistically significantly different from intrasurgical measurements (p < 0.005).

of a full-thickness flap, the distances from the cementoenamel junction (CEJ) to alveolar crest (AC) and from the CEJ to the most apical extension of the bony defect (BD) were measured to the nearest 0.5 mm using a PCPUNC 15-probe. All clinical measurements were performed by one examiner (RE.). Reliability of PD and PAL measurements has been demonstrated previously for molars. 16

Radiographic evaluation. Anatomic landmarks. The BD was defined as the most coronal point where the periodontal ligament space showed a continuous width (Fig. 1). If no periodontal ligament space could be identified, the point where the projection of the AC crossed the root surface was taken as the landmark. 11 If both structures could be identified at one defect, the point defined by the periodontal ligament was used as the BD, and the crossing of the silhouette of the alveolar crest with the root surface was defined as the AC. If several bony contours could be identified, the most apical that crossed the root was defined as the BD and the most coronal as the AC (Fig. 2). 6 Loupe method. The distance between the projections of the orthodontic wires that had been fixed to the filmholders were measured vertically (dr) and horizontally (dh) on every radiograph using a loupe of 10-fold magnification and a 0.1 m m grid (Scale loupe 10×, Peak, Tohkai Sangyo, Tokyo, Japan). 14 The length of the cast shadow of the wire placed on the maxillary side

of the filmholder was measured in the same manner, and the mean enlargement of the radiographs was calculated. 6 Within the interproximal defect, the linear distances between the landmarks CEJ and AC as well as CEJ and BD were assessed to the nearest 0.1 mm. Digital method. All radiographs were captured using a video camera (WV-BD 400, Panasonic, Secaucus, N.J.) and after digitization transferred to a computer. The program used to measure linear distances was LinearMeaSuRemenT (LMSRT, University of Florida, Gainesville, Fla.). 1°,11 Under 18-fold magnification, a region of interest (ROI), measuring 256 x 256 pixels was selected. The landmarks CEJ, AC, and BD were marked on the screen. After marking the reference points, the picture was stored and the ROI, including the reference points, was displayed in another window on the screen. The selection of the landmarks was then repeated on the original image. The program calculated the measured distances for first and repeated measurements, the average length and measurement error, ll Eighteen radiographs were analyzed by one examiner (RE.) and 44 by another (T.S.K.) using the loupe and LMSRT. All radiographic measurements were adjusted according to the mean enlargement of the radiographs.

Biometric evaluation. To estimate the validity of loupe and computer-assisted measurements, the distances that had been measured on the presurgical radiographs were compared to the intrasurgical assessments used as gold standard. 4,6 For

102 Eickholz et al.

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY January 1998

30 T

I. . . . . . .

[

25

20

n

15 1o

5

0 <.-4.0

-4.0-(-3.1)

-3.0-(-2.1)

-2.0-(-1,1)

-1.0-(-0.1)

0-0.9

1.0-1,9

2,0-2.9

>3.0

mm

Fig. 3. Distribution of differences loupe measurement minus intrasurgical measurement (loupe) and LMSRT measurement minus intrasurgical measurement (LMSRT) for the distance CEJ to AC (n: number of sites).

30

25

20

El 1 5

10

5

0 <-5,0

5.0~4.1)

4.0-(-3.1)

-3.0~2.1)

-2.0-(-1.1)

-1.0-(0.1)

0-0.9

1.0-1.9

2.0-2.9

>3.0 m m

Fig. 4. Distribution of differences loupe measurement minus intrasurgical measurement (loupe) and LMSRT measurement minus intrasurgical measurement (LMSRT) for the distance CEJ to BD (n: number of sites).

Table III. Validity of the assessment of linear distances on radiographs (radiographic measurement) using a loupe and LMSRT compared to the gold standard of intrasurgical measurements Difference: radiographic-intrasurgicaI measurement Distances

Loupe (mm)

LMSRT (mm)

CEJ-AC CEJ-BD

- 0 . 8 6 + 2.09 - 1 . 2 2 _+ 2.33

- 0 . 5 3 + 1.86" - 0 . 8 0 + 2.09*

*Statistically significantly different from measurement by loupe (p < 0.005).

all defects, two intrasurgical assessments were performed at the interproximal defects: a mesiobuccal and mesiolingual or a distobuccal and distolingual measurement. For comparison with the radiographic measurements, the higher score of both measurements was always used. For each defect, the difference between intrasurgical assessment and measurement with the

loupe, as well as intrasurgical assessment and computer-assisted measured distance, was calculated for the distances from CEJ to AC and from CEJ to BD. After testing the data for normality using the KolmogorovSmirnov test and Lilliefors test, the means of the loupe and the LMSRT measurements for the distances CEJAC and CEJ-BD were compared to the respective intrasurgical measurements using a paired t-test. Further, the means of the differences between intrasurgical and loupe assessments were compared with the differences between intrasurgical and LMSRT measurements using a paired t-test. Because of multiple testing, a correction according to Bonferroni was performed. A stepwise multiple linear regression analysis was applied to identify factors (angulation difference, examiner, patient, GI, PII, PD, PAL, CEJ-AC, CEJ-BD) that might influence the validity (differences radiographic minus intrasurgical CEJ-AC/CEJ-BD) of the more valid analysis. The examiner was defined by dummy variables. A probability o f p < 0.1 was used to retain a variable in the

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY

Eickholz et al.

103

Volume 85, Number 1

Table IV. Stepwise multiple linear regression analysis Dependent variable: A radiographic-intrasurgicaI ( CEJ-AC); n = l l & R 2 = 0.970; R2adjusted = 0.968; S.E.(est.) = 0.320

Constant Horizontal angulation Vertical angulation CEJ - AC (LMSRT) CEJ - AC (intrasurgical)

Coefficient

SE (b)

-0.052 0.092 0.036 0.937 -0.976

0.100 0.044 0.020 0.018 0.020

Tolerance

p

0.934 0.940 0.783 0.780

0.601 0.040 0.071 0.000 0.000

Analysis of variance Source Regression Residual

SSQ

DF

MSQ

F-ratio

p

359.366 ' 11.276

4 110

89.841 0.103

876.422

0.000

Table V. Stepwise multiple linear regression analysis Dependent variable: A radiographic-intrasurgical (CEJ-BD); n = 115," R 2 = 0.965," R2adjusted= 0.964; s.c.(est.) = 0.405

Constant Horizontal angulation CEJ - BD (LMSRT) CEJ - BD (intrasurgical)

Coefficient

SE (b)

Tolerance

p

-0.024 0.094 1.009 -0.993

0.119 0.054 0.018 0.021

0.992 0.783 0.515

0.841 0.086 0.000 0.000

Analysis of variance source regression residual

SSQ

DF

MSQ

F-ratio

p

495.873 18.212

3 111

165.291 0.164

1007.410

0.000

model. Statistical analysis was performed using Systat for Windows version 5.03 (Systat Inc, Evanston, Ill.). RESULTS Sixty-two standardized radiographs, with 115 interproximal periodontal lesions, were obtained. The clinical parameters of the interproximal defects and the angulation difference of the central beam from the orthoradial projection are given in Table I. The mean magnification _+ standard deviation of the radiologic images was 1.03 + 0.02. The results of the intrasurgical measurements and the radiographic assessments using the loupe and LMSRT are shown in Table II. Both analyzing techniques statistically significantly underestimated the amount of interproximal bone loss compared with intrasurgical measurements. Using LMSRT, a statistically significant (p < 0.001) smaller underestimation of bone loss was observed compared with loupe measurements (Table III). The distribution of the differences between loupe and intrasurgical as well as LMSRT and intrasurgical measurements is given in Fig. 3 for the distance CEJ to

AC and in Fig. 4 for the distance CEJ to BD. The discrepancy between LMSRT and intrasurgical assessments was modulated by factors of vertical and horizontal angulation difference from the orthoradial projection as well as defect depth, whereas the different examiners and patients did not influence the validity of the radiographic measurements (Tables IV and V). The particular models explained 97% (CEJ-AC) and 96% (CEJ-BD) of the variation of the dependent variable.

DISCUSSION Alveolar bone loss attributable to periodontitis can be assessed using intraoral radiographs; however, these radiographs provide only two-dimensional images of three-dimensional structures. Changes in the projection geometry between consecutively obtained radiographs may lead to different two-dimensional images of the same three-dimensional situation. Such artifacts often are difficult to distinguish from biologic changes. Hence, the projection geometry of serial radiographs has to be highly standardized. Projection artifacts may be caused by different angulations between the central

104

EickhoIz et al.

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY

January 1998 beam related to the filmholder and the film while the relation between teeth and film is fixed. 17 Another source of projection artifacts is angulation differences between the central beam and the anatomic structures to be imaged. 18 Prefabricated filmholders may provide projection standardization to a certain degree. 19,2° The filmholder position may be standardized using impressions that are supported unilaterally7,14,21-24 or bilaterally. 18,25,26 If the radiation source is coupled to the filmholder with definition of the relation between filmholder and teeth by an impression, 8,18,22,23 this rigid connection may cause tilting between impression and teeth that is difficult to control. 1° Thus, other authors favor the uncoupled alignment of the central beam. 7,14,21,24 With the use of uncoupled alignment devices, small angulation differences are likely to occur. Hence, it is important to have information about projection geometry. 7,14,24 As previously reported by other authors, we observed mean vertical angulation differences between the central beam and the orthoradial projection of 1.0 to 1.6 degrees; these were approximately twice as high as the horizontal angulation differences 0.6 to 0.9 degrees. 24,26,27 In a previous study, it was shown that for the standardization technique used in this investigation, the angulation difference between teeth and filmholder, for consecutive radiographs obtained during a 3-month interval, was smaller than the measurement error of the assessments needed for the calculation of these angulations. 14 Hence, it was concluded that the angulation difference between central beam and filmholder is more important for the interpretation of radiographic changes than the angulation between filmholder and teeth. 14 In one study, measurements of linear distances were made between endodontic landmarks (coronal reference, most apical extension of root canal filling, radiographic apex) on master cone trial and control radiographs obtained during endodontic therapy for about 30 minutes. 28 No statistically significant differences were found between the respective distances on master cone trial and control radiographs, respectively.28 This is further evidence to support previous findings that there is only minor influence of discrepancies between tooth and film or filmholder on the standardization obtained with the alignment technique used. This method seems to be sufficient to avoid projection artifacts when measuring linear distances. The high reproducibility of the preceding technique allowed the assessment of even small bony changes. 6 Estimating the intraindividual reproducibility of measuring linear distances between periodontal landmarks (CEJ-AC, CEJ-BD) to the nearest 0.1 mm using a calibrated loupe, a standard deviation of single measurements of 0.25 mm was calculated. 14

Comparing conventional evaluation of radiographs using a loupe with 10-fold magnifcation and a calibrated grid in 0.1 increments and a computer-assisted analysis (18-fold magnification) with the gold standard of intrasurgical measurements, the computer-assisted analysis was more valid, although both methods underestimated the amount of interproximal bone loss. Using the loupe and measuring to the nearest 0.1 mm, the distance from CEJ to BD was underestimated by 1.2 _ 2.3 mm, and for LMSRT underestimation was 0.8 _ 2.1 mm. When evaluating untreated interproximal intrabony defects, Tonetti et al. 5 underestimated intrasurgically measured bone loss by 1.3 mm (CEJ-BD) and 0.6 m m (CEJ-AC) using a computer-assisted device. /~kesson et al.4 underestimated intrasurgically assessed interproximal bone loss by 2.3 _+2.0 mm when measuring radiographic bony lesions to the nearest 0.1 m m using a caliper. The computer-assisted analysis underestimation of bone loss was statistically significantly smaller. These observations confmn data reported previously after evaluation of a smaller sample by only one examiner. 6 Shrout et al. 3 underestimated overall bone loss by only 0.58 _+0.75 mm by measuring mesial, midtooth, and distal sites on computer-enhanced radiographs of dried skulls. The computer-assisted enhancement may account for these differences. However, these authors introduced no scattering equivalent to their experimental study. In clinical investigations, x-ray scattering is more pronounced by soft tissues than in radiographs of dried skulls. Thus, the quality of these radiographs may be better and, therefore, easier to measure than clinical radiographs of periodontitis patients. Shrout et al. 3 also did not adjust the radiographi c measurements according to the enlargement of the radiographs. The radiographic enlargement may partially have compensated for the underestimation of bone loss by radiographic measurements. The implementation of a double measurement that is averaged in the computer-assisted analysis may reduce measurement errors as compared to the loupe method. The loupe measurements were performed under 10-fold magnification to the nearest 0.1 mm, whereas the computer-assisted measurements were performed under 18fold magnification with a higher resolution. These differences may account for the higher validity of the LMSRT. After it was revealed that radiographic measurements with LMSRT were more valid than with loupe, further analysis was limited to the superior method. By multiple linear regression analysis, the length of computerassisted and intrasurgicaUy assessed distances CEJ to AC and CEJ to BD, respectively, were identified as factors influencing the discrepancy between distances measured by LMSRT or intrasurgically: the larger the

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY

Eickholz et al. 105

Volume 85, Number 1

radiographic measurement, the smaller the underestimation of alveolar crest or apical extension of the bony defect. The larger the intrasurgical assessment, the larger the underestimation of the gold standard; that is, the deeper the defect, the more likely it is to be underestimated by radiographic measurements. This observation is in contrast to findings of Hgmmerle et al., 29 who found radiographic measurements to overestimate advanced bone loss and to underestimate superficial bone loss. Differences in the definition of the reference points for the radiographic measurements may account for these differences. When two different reference points were found in this study, the more coronally located landmark was taken as AC and the more apically reference point was taken as BD, whereas Hfimmerle et al. 29 did not distinguish between the landmarks AC and BD and calculated a mean value of the two measurements. Further these authors did not account for the influence of angulation between central beam and orthoradial projection. Vertical and horizontal angulation difference between central beam and orthoradial projection influenced the radiographic measurement of the distance CEJ to AC, whereas just the horizontal angulation influenced the discrepancy between radiographic measurement of the distance CEJ to BD and gold standard. The larger the horizontal angulation difference, the larger the underestimation of bone loss (distance CEJ-BD) by radiographic measurement. This seems to be plausible because under strong horizontal angulation of the central beam to the orthoradial projection, there is a high risk that the root of a neighboring tooth would be projected on the interproximal space and thereby obscure the intrabony defect. Thus, the intrabony defect would not be detected. Vertical angulation to the orthoradial projection may cause a double silhouette of the alveolar crest. Aiming the central beam from an apical direction would project the buccal margin of the alveolar crest more coronally in relation to the CEJ; thus, the distance from CEJ to AC would likely be underestimated. Compared to the influence of the factors intrasurgical and radiographic measurement, however, the influence of angulation deviation from orthoradial projection is moderate. This confirms observations of Shrout et al.3 who reported no substantial influence on the validity of radiographic measurements of beam positioning errors up to 10 degrees in an experimental study. Another factor influencing radiographic measurement error revealed by Tonetti et al.5 is the disease status of a lesion. These authors found a higher underestimation of bone loss in untreated than in treated sites. Interestingly, the use of different examiners did not influence the validity of computer-assisted radiographic measurements. It seems that at least for well-trained

and calibrated examiners, the LMSRT analysis does not depend on examiner characteristics in a significant degree. Our results confirm the importance of an orthoradial projection for the radiographic examination of interproximal bone loss. By analyzing 115 sites o n 62 radiographs of 35 patients, multiple measurements per patient and per radiograph were performed. For our analysis, we assumed measurements to be independent from each other. However, patient-related factors (e.g., individual bone density) and radiograph-related factors (e.g., as a result of overexposition or underexposition) may have influenced the dependent variable. Thus, only correctly exposed radiographs of good quality were analyzed. Further, in the present study, a radiographrelated factor has been shown to influence the validity of radiographic measurements (i.e., angulation between central beam and orthoradial projection). For this sample, however, the individual patient could not be identified as a factor influencing the difference between L M S R T and intrasurgical measurements of the distances CEJ to AC and CEJ to BD (i.e., individual measurements may be looked on as independent). It seems that the evaluation of radiographs that takes place outside the oral cavity is less influenced by patient factors than the clinical evaluation of progressing disease or therapeutic changes. Within the limits of the present study, the following conclusions can be drawn: (1) the computer-assisted analysis of linear distances on radiographs underestimated the amount of interproximal bone loss statistically significantly less than conventional measurements using a calibrated loupe and (2) vertical and particularly horizontal angulation differences between the central beam and the orthoradial projection increase the risk of underestimating interproximal bone loss by radiographic examination. REFERENCES

1. Theilade J. An evaluation of the reliability of radiographs in the measurement of bone loss in periodontal disease. J Periodontol 1960;31:143-53. 2. Suomi JD, Plumbo J, Barbano JE A comparative study of radiographs and pocket measurements in periodontal disease evaluation. J Periodontol 1968;39:311-5. 3. Shrout MK, Hildebolt CE Vannier MW. The effect of alignment errors on bitewing-based bone loss measurements. J Clin Periodontol 1991;18:708-12. 4. ~&esson L, Hhkansson J, Rohlin M. Comparison of panoramic and intraoral radiography and pocket probing for the measurement 0f marginal bone level. J Clin Periodontol 1992;19:326-32. 5. TonettiMS, Pini Prato G, Williams RC, Cortellini R Periodontal regeneration of human infrabony defects. II1. Diagnostic strategies to detect bone gain. J Periodontol 1993;64:269-77. 6. Eickholz E Benn DK, Staehle HJ. Radiographic evaluation of bone regeneration followingperiodontal surgery with or without expanded polytetrafluoroethylenebarriers. J Periodontol 1996; 67:379-85. 7. DuckworthJE, Judy PF, Goodson JM, SocranskySS. A method

106 Eickholz et al.

ORAL SURGERY ORAL MEDICINE ORAL PATHOLOGY

January 1998

8. 9. 10. 11. 12.

13.

14. 15. 16. 17. 18.

19. 20.

for the geometric and densitometric standardisation of intraoral radiographs. J Periodontol 1983;54:435-40. Jeffcoat MK, Wang I-C, Reddy MS. Radiographic diagnosis in periodontics. Periodontology 2000 1995;7:54-68. van der Stelt PF. Modem radiographic methods in the diagnosis of periodontal disease. Adv Dent Res 1993;7:158-62. Benn DK. A review of the reliability of radiographic measurements in estimating alveolar bone changes. J Clin Periodontol 1990; 17:14-21. Benn DK. A computer-assisted method for making linear radiographic measurements using stored regions of interest. J Clin Periodontol 1992;19:441-8. Hansmann E, Christersson L, Dunford R, Wikesjo U, Phyo J, Genco RJ. Usefulness of subtraction radiography in the evaluation of periodontal therapy. J Periodontol 1985;56(supplement):4-7. Hansmann E, Allen KM, Piedmonte MR. Influence of variations in projection geometry and lesion size on detection of computer-simulated crestal alveolar bone lesions by subtraction radiography. J Periodont Res 1991 ;26:48-51. Eickholz P, D6rfer C, Staehle HJ. Reproduzierbarkeit standardisierter Bi~fltigelanfnahmen bei Patienten mit fortgeschrittener Parodontitis. Dtsch Zahn~rztl Z 1994;49:398-402. L6e H. The gingival index, the plaque index and the retention index system. J Periodontol 1967;38:610-6. Eickholz P. The reproducibility and validity of furcation measurements as related to class of farcation invasion. J Periodontol 1995;66:984-9. Sewerin IB, Andersen V, Stoltze K. Influence of projection angles upon position of cementoenamel junction on radiographs, Scand J Dent Res 1987;95:74-81. Zappa U, Simona C, Graf H, van Aken J. In vivo determination of radiographic projection errors produced by a novel filmholder and an x-ray beam manipulator. J Periodontol 1991;62:67483. Eggen S. Standardized projection technique in intraoral roentgenography. Sveriges Tandl~ikarf6rbunds Tidning 1969; 61:867-72. Updegrave WJ. The paralleling extension-cone technique in

21.

22. 23. 24. 25.

26. 27. 28. 29.

intraoral dental radiography. Oral Surg Oral Med Oral Pathol 1951 ;4:1250-61. Bragger U, Pasquali L, Kornman KS. Remodelling of interdental alveolar bone after periodontal flap procedures assessed by means of computer-assisted densitometric image analysis (CADIA). J Clin Periodontol 1988; 15:558-64. Dubrez B, Jacot-Descombes S, Cimasoni G. Reliability of a paralleling instrument for dental radiographs. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995;80:358-64. Payot P, Haroutunian B, Pochon Y, Herr P, Bickel M, Cimasoni G. Densitometric analysis of lower molar interradicnlar areas in superposable radiographs. J Clinical Periodontol 1987; 14:1-7. Schulz A, Mtiller H-P, Topoll HH, Lange DE. Abweichungen des Zentralstrahls bei standardisierten Bi[3fliigelaufnahmen. Dtsch Zahn~ztl Z 1991;46:505-8. Janssen PTM, van Palenstein Helderman WH, van Aken J. The effect of in vivo-occurring errors in the reproducibility of radiographs on the use of the subtraction technique. J Clin Periodontol 1989;16:53-8. Potter BJ, Shrout MK, Harrell JC. Reproducibility of beam alignment using different bite-wing radiographic techniques. Oral Surg Oral Med Oral Pathol 1995;79:532-5. Shrout MK, Hildebolt CF, Vannier MW. Alignment errors in bitewing radiographs using uncoupled positioning devices. Dentomaxillofac Radiol 1993;22:33-7. Eickholz P, Miiller H-P. Reproduzierbarkeit standardisierter R6ntgenaufnahmen in der endodontischen Diagnostik. Dtsch Zahn~ztl 1994;Z 49:796-800. H~miaerle CHF, Ingold H-P, Lang NP. Evaluation of clinical and radiographic scoring methods before and after initial periodontal therapy. J Clin Periodontol 1990;17:255-63.

Reprint requests: Dr. Peter Eickholz Poliklinik ftir Zahnerhaltungskunde Im Neuenheimer Feld 400 69120 Heidelberg, Germany