30149.

qxd 8/11/04 7:23 AM Page 1145

J Periodontol • August 2004

Comparison Between Subtraction

Radiography and Conventional
Radiographic Interpretation During
Long-Term Evaluation of Periodontal
Therapy in Class II Furcation Defects
Patricia R. Cury,* Ney S. Araujo,† Jon Bowie,‡ Enilson A. Sallum,§ and Marjorie K. Jeffcoat

Background: Limited information comparing digital subtraction radi-

ographic assessment with conventional radiographic interpretation is
available from longitudinal clinical trials. The aim of this study was to
evaluate the ability to detect periodontal bone changes during the
long-term maintenance of Class II furcation defects by conventional
radiographic interpretation compared to interpretation of digital sub-
traction images.

C
onventional radiographs have
Methods: Standardized radiographs of 18 Class II furcation defects
been the primary diagnostic
in mandibular molars were taken at baseline and at 6, 12, 18, and
method for the detection of
24 months after non-resective periodontal surgery. Conventional
osseous changes due to periodontitis
radiographic and digital subtraction interpretations were performed
masked, respectively, by two and three experienced examiners, or periodontal treatment. They are
according to the following categories: bone gain; bone loss; un- also used for the longitudinal follow-
changed appearance; and impossible to visualize. Percent concor- up of periodontal patients. However,
dance and the kappa statistic value (κ) were computed. low sensitivity1 in detecting a small
Results: Conventional radiographic and digital subtraction interpre- osseous lesion and the high intra- and
tation images resulted in 72 decisions for each examiner. The visual inter- interexaminer variability2 may limit
pretation of digital subtraction images by two examiners revealed the their diagnostic value.
same results. The interpretation of conventional radiographic images Due to elimination of all the ana-
showed a low concordance between examiners (κ <0.40) at all exam- tomical structures other than those of
inations. The concordance between subtraction radiography and conven- immediate interest, subtraction radi-
tional radiographic interpretation was also low for all examiners (κ <0.36) ography has shown a high sensitivity
at all examinations. Using subtraction radiography as a reference, bone for the detection of subtle bone
changed and bone unchanged were diagnosed correctly in 47.2% of changes.3 Many studies on dry bone
cases by examiner A, in 43.1% by examiner B, and in 38.9% by exam- specimens comparing digital sub-
iner C. traction radiographic assessment and
Conclusion: It can be concluded that conventional radiographic conventional radiographic interpreta-
interpretation is a more subjective and inaccurate method of detect- tion have shown that subtraction radi-
ing periodontal bone changes in Class II furcation defects in mandibu- ography significantly increases the
lar molars when compared with subtraction radiography. J Periodontol detectability of small alveolar bone
2004;75:1145-1149. changes.1,4-7 However, these studies
KEY WORDS were performed in vitro and clinical
situations involve more difficult stan-
Bone loss; clinical trials; follow-up studies; furcation/diagnosis;
dardization of radiographic images
radiography, dental; radiography, subtraction.
and the presence of soft tissues.
Limited information comparing
* Department of Microbiology, São Leopoldo-Mandic Dental Research Institute, Campinas, Brazil. interpretation of subtraction radio-
† Department of Oral Pathology, University of São Paulo, School of Dentistry, São Paulo, Brazil.
‡ Private practice, Fairhope, AL. graphs with conventional radiographs
§ Department of Prosthodontics and Periodontics, Division of Periodontics, School of Dentistry at is available from clinical studies in
Piracicaba, State University of Campinas, São Paulo, Brazil.

Department of Periodontics, School of Dental Medicine, University of Pennsylvania, Philadelphia, humans.2,8,9 All of these studies
PA. showed that subtraction radiography

1145
30149.qxd 8/11/04 7:23 AM Page 1146
Subtraction Radiography Versus Conventional Radiographs
was superior to conventional radiography in detecting
bone changes. However, these studies involved only inter-
dental sites8 or did not describe the type of defect stud-
ied.2,9 The aim of this study was to compare conventional
radiographic interpretation with interpretation of digital
subtraction images to detect periodontal bone changes
during long-term maintenance of Class II furcation defects.
MATERIALS AND METHODS
The study protocol was approved by the University of
Campinas Institutional Committee of Ethics. Informed
consent was obtained from all subjects. Eighteen Class II
furcation defects in lower molars of nine systemically
healthy patients diagnosed with chronic periodontitis were
selected for study. The defects were treated by either
guided tissue regeneration¶ or open flap debridement as
previously described.10
Standardized vertical bitewing radiographs# were
obtained by one periodontist at baseline (immediately
before surgery) and at 6, 12, 18, and 24 months after
periodontal surgery. The radiographs were taken with
the long-cone paralleling technique employing an x-ray
unit operating at 70 kV, 10 mA, and 0.8 seconds expo-
sure time. In order to standardize the geometry, a mod-
ification of the system** connected to an individual
acrylic bite-block with occlusal registration of autopoly-
merizing acrylic resin was used. The films were developed
by an automatic film processor following the manu-
facturer’s instructions.††
The radiographs for 18 molars from four follow-up
periods were used, for a total of 72 subtraction radi-
ography and conventional radiographic evaluations for
each examiner. Five periodontists were informed that
the films were from pre- and post-operative radiographs
of periodontal surgeries and directed to score the visual
changes in the furcation defects according to the fol-
lowing categories: bone gain; bone loss; unchanged
appearance; impossible to visualize. Conventional radi-
ographic and digital subtraction interpretations were
performed masked, respectively, by three and two experi-
enced examiners. The examiners were not aware of
the surgery performed or the clinical results. All exam-
iners were experienced periodontists trained in inter-
preting conventional radiographs and/or subtraction
radiography.
Conventional Radiographic Evaluation
For the conventional radiographic interpretation, the
examiners were asked to observe the baseline and
follow-up films, mounted in pairs in non-transparent
frames, on a masked view box. The pairs of radi-
ographs were shown in a randomized order. The read-
ings were divided into four sessions separated by at least
2 weeks to avoid fatigue and to ensure that images of
the same site and period were not viewed at the same
occasion.
1146
Volume 75 • Number 8
Subtraction Radiography Analysis
For each site, the baseline radiograph was placed under
a video camera, digitized with 512 × 480 pixels of space
resolution and eight bits (256 gray levels) of color res-
olution and stored in the computer. The follow-up radi-
ographs were aligned under the video camera and
digitized. Pairs of radiographs were subtracted follow-
ing correction for contrast and planar geometric dis-
crepancies as previously described.3 The resultant
subtraction image showed areas of bone loss (dark
areas) and bone gain (light areas) against a neutral gray
background.
Areas of bone loss or gain were isolated using a
morphologically aided technique, which also removes
background noise from the image. The signed sub-
traction image was converted to a binary image (black
and white with no shades of gray) using an interactively
controlled threshold. The operator adjusted the thresh-
old until the area of bony change appeared white
against a background of “salt and pepper” noise. An
erode operation was performed on the binary image
to aid in removing the isolated pixel noise. A dilate
operation was then performed to restore the areas of
bony change to their original size. This erode-dilate
operation is known as a morphological OPEN.
Statistical Analysis
To compare the concordance between subtraction radi-
ography and conventional radiographic interpretation,
as well as the concordance between examiners, per-
centage of concordance and the kappa statistic (κ)
were computed. The overall agreement, that is, the
total number of sites on which observers agree regard-
less of whether they agree that changes have occurred,
was also calculated.
RESULTS
Seventy-two comparisons between conventional radi-
ographic interpretation and subtraction radiography
were computed for three and two examiners, respec-
tively, and also between examiners of conventional
radiographs.
The visual interpretation of digital subtraction images
was an objective examination and the interpretation by
two examiners revealed the same results. The inter-
pretation of conventional radiographs showed a low
concordance between examiners (κ ≤0.40) at 6, 12,
18, and 24 months (Table 1). However, 58.33% of
kappa coefficients were significantly greater than zero
(P <0.05).
The concordance between subtraction radiography
and conventional radiographic interpretation was low for
all examiners at all examinations (κ ≤0.36; Table 2).
¶ Guidor bioresorbable matrix barrier, Guidor AB, Novum, Huddinge, Sweden.
# Kodak EctaSpeed dental films, Eastman Kodak Company, Rochester, NY.
** XCP, Rinn Corporation, Elgin, IL.
†† GXP, Gendex, New York, NY.
30149.qxd 8/11/04 7:23 AM Page 1147

J Periodontol • August 2004 Cury, Araujo, Bowie, Sallum, Jeffcoat

Table 1. examiner A detected eight (47.1%), examiner B detected

10 (58.8%), and examiner C detected four (23.5%).
κ) and P Values
Kappa Coefficients (κ
Fourteen digital subtraction images were not accept-
for Examiner Agreement on Conventional able for analysis due to anatomic structures overlying or
Radiographs at 6, 12, 18, and 24 Months radiographs with poor quality. Otherwise, for conventional
and the Average Between Examiners radiographic interpretation, examiner A could not ana-
(N = 72) lyze nine of these 14 radiographs (64.3%), examiner B,
seven (50%), and examiner C, three (21.4%).
Examiner A Examiner B
DISCUSSION
κ P κ P The present study compared conventional radiographic
6 months Examiner B 0.09 0.44 interpretation with a validated subtraction program.3,11-13
Examiner C 0.22 0.11 0.40 0.01 A previous study has shown that sensitivity and speci-
ficity in detecting bone chips less than 10 mg in mass
12 months Examiner B 0.39 0.01 fixed on the mandible were between 89% and 100%,12
Examiner C 0.39 0.01 0.21 0.01 and for 1 mg chips, 87.8% and 100%.3 Also, there was
18 months Examiner B 0.29 0.02 a strong linear relationship between actual lesion mass
Examiner C 0.15 0.23 0.30 0.01 and calculated mass (R2 = 0.94, slope = 0.98,
P <0.0001),3 which demonstrates the accuracy of the
24 months Examiner B 0.33 0.02
software program. Therefore, in the present study, this
Examiner C 0.09 0.39 0.19 0.06
validated subtraction software was considered the ref-
Mean kappa Examiner B 0.265 0.01 erence to evaluate the ability to detect bone changes
score by conventional radiographic interpretation.
Mean kappa Examiner C 0.206 0.01 0.31 0.01
In the present study, there was substantial dis-
score agreement between conventional and subtraction radi-
ography interpretation and also between examiners of
the conventional radiographs,
which agrees with the litera-
Table 2.
ture.2,8,9 However, these stud-
Kappa Coefficients (κ κ) and P Values for Agreement Between ies, in contrast to the present
Subtraction Radiography and Conventional Radiographic study, involved only interdental
sites 8 and some sites received
Interpretation at 6, 12, 18, and 24 Months (N = 72) for Each
ostectomy,8,9 which could make
Examiner the conventional radiographic
evaluation easier. Also, the time
Subtraction Radiography intervals between the radi-
Examiners ographic examinations were dif-
6 Months 12 Months 18 Months 24 Months
(conventional Mean ferent, from 2 to 6 weeks8,9 to
radiography) κ P κ P κ P κ P κ Score P 56 months.2 Bragger et al.8
Examiner A 0.11 0.41 0.32 0.02 0.34 0.02 0.32 0.01 0.27 0.01 showed that the conventional
interpretation of radiographs
Examiner B 0.12 0.35 0.32 0.02 0.36 0.01 0.21 0.06 0.26 0.01 detected only 50% of the sites
Examiner C 0.08 0.55 0.14 0.26 −0.08 0.56 0.06 0.67 0.08 0.28 with bone loss, while in the pre-
sent study, 25% to 65.7% of
sites with bone loss were diag-
nosed correctly.
Using subtraction radiography as a reference, bone Hausmann et al.14 and Wenzel et al.15 also analyzed
conditions were diagnosed correctly in 47.2% of the cases a set of radiographs by subtraction radiography and
by examiner A, in 43.1% by examiner B, and in 38.9% by measurement of alveolar bone crest height, show-
by examiner C (Table 3). From nine diagnoses of bone ing that subtraction radiography was more sensitive in
gain revealed by subtraction radiography, examiner A detecting change, and their results agree with our study.
detected only one (11.1%), examiner B detected six Wenzel et al.15 showed bone gain in 56% and 22% of
(66.7%), and examiner C did not detect any bone gain. the sites after 6 months and 44% and 66% after 12
From 32 diagnoses of bone loss, examiner A detected months by conventional radiography and digital sub-
16 sites (50%), B detected eight (25%), and C detected traction radiography, respectively. However, in contrast
21 (65.7%). From 17 diagnoses of unchanged condition, with the present study, for the conventional radiographic

1147
30149.qxd 8/11/04 7:23 AM Page 1148
Subtraction Radiography Versus Conventional Radiographs
Table 3.
Sites Presenting Bone Gain, Bone Loss,
Unchanged, and Impossible to Analyze
According to Subtraction Radiography
and Agreement (%) Between Each
Examiner for Conventional and Subtraction
Radiography (N = 72)
Overall
Agree-
Bone Gain Bone Loss Unchanged Impossible ment
Examiner (N = 9) (N = 32) (N = 17) (N = 14) (%)
A 11.11 50.00 47.06 64.29 47.22
B 66.67 25.00 58.82 50.00 43.06
C 0.00 65.66 23.53 21.43 38.89
analysis, bone-height measurements were performed on
the video display, with seven times magnification on the
monitor screen14 or projected on white paper, at ×15.3
magnification.15
The conventional radiographic interpretation results
achieved in our study and others2,8,9,14,15 may be supe-
rior to that in a private practice, where the individual acrylic
bite-block with occlusal registration of autopolymerizing
acrylic resin or magnification are not usually used.
In this study the interpretation of conventional radi-
ographs showed a low concordance between examiners
(κ ≤0.40). Grondahl et al.,2 however, found that in only
a few cases, the observers reported observations quite
contrary to each other.
One disadvantage of digital subtraction radiogra-
phy, as used presently, is the need for nearly identical
projection alignment during the exposure of the sequen-
tial radiographs, using customized occlusal stents10,16
or a cephalostat.13 The occlusal stents made of cold-
cured acrylic or impression materials align the films to
the dentition and are feasible for a limited number of
patients and follow-up periods that do not exceed 2 years.
To alleviate this problem, Jeffcoat et al. introduced a
method that combines a cephalometric head posi-
tioning device and a long source-to-object distance.13
However, this method is not feasible in most conven-
tional dental practices due to the high-cost specialized
positioning equipment. Furthermore, highly special-
ized computer image processing equipment and soft-
ware are required for image analyses. Perhaps, in
clinical practice, most of the periodontal evaluations
still rely on conventional radiographic interpretation
because of these drawbacks.
Unlike those studies3,13 in which the radiographic
images were acquired under strictly controlled head
positioning and exposure, diagnostic subtraction radi-
1148
Volume 75 • Number 8
ography (DSR), which combines the use of a position-
ing device during film exposure with specialized software
for digital image subtraction, appears to be practical
under the conditions found in most dental practices.9
Comparison of DSR to the established and highly stan-
dardized cephalostat-guided image acquisition method
showed that there was no significant difference between
these two methods.9 The exposures were performed
with a conventional dental x-ray unit and the film was
positioned using the DSR aiming device equipped with
an additional beam collimator support.
Radiographic examination is essential to detect
osseous changes caused by periodontitis and to eval-
uate the long-term outcome of periodontal therapy.
However, considering the substantial disagreement
between subtraction radiography and conventional radi-
ographic interpretation (κ <0.36), and even between the
examiners (κ <0.40), patients may not have been
receiving optimum therapy due to misinterpretation of
conventional radiographs.
Within the limits of the present study, it can be con-
cluded that conventional radiographic interpretation is
a more subjective and inaccurate method of detecting
periodontal bone changes in Class II furcation defects
in lower molars compared to subtraction radiography.
REFERENCES
1. Grondahl K, Kullendorff B, Strid KG, Grondahl HG,
Henrikson CO. Detectability of artificial marginal bone
lesions as a function of lesion depth. A comparison
between subtraction radiography and conventional radi-
ographic technique. J Clin Periodontol 1988;15:156-162.
2. Grondahl K, Grondahl HG, Wennström J, Heijl L. Exam-
iner agreement in estimating changes in periodontal bone
from conventional and subtraction radiographs. J Clin
Periodontol 1987;14:74-79.
3. Jeffcoat MK, Reddy MS, Magnusson I, et al. Efficacy of
quantitative digital subtraction radiography using radi-
ographs exposed in a multicenter trial. J Periodontal Res
1996;31:157-160.
4. Janssen PT, van Palenstein Helderman WH, van Aken J.
The detection of in vitro produced periodontal bone
lesions by conventional radiography and photographic
subtraction radiography using observers and quantitative
digital subtraction radiography. J Clin Periodontol 1989;
16:335-341.
5. Grondahl K, Grondahl HG, Webber RL. Influence of vari-
ations in projection geometry on the detectability of peri-
odontal bone lesions. A comparison between subtraction
radiography and conventional radiographic technique.
J Clin Periodontol 1984;11:411-420.
6. Kullendorff B, Grondahl K, Rohlin M, Nilsson M. Subtrac-
tion radiography of interradicular bone lesions. Acta
Odontol Scand 1992;50:259-267.
7. Nicopoulou-Karayianni K, Bragger U, Burgin W, Nielsen PM,
Lang NP. Diagnosis of alveolar bone changes with digital sub-
traction images and conventional radiographs. An in vitro
study. Oral Surg Oral Med Oral Pathol 1991;72:251-256.
8. Bragger U, Pasquali L, Rylander H, Carnes D, Kornman
KS. Computer-assisted densitometric image analysis in
periodontal radiography. A methodological study. J Clin
Periodontol 1988;15:27-37.
30149.qxd 8/11/04 7:23 AM Page 1149

J Periodontol • August 2004 Cury, Araujo, Bowie, Sallum, Jeffcoat

9. Nummikoski PV, Steffensen B, Hamilton K, Dove SB. 16. McHenry K, Hausmann E, Wikesjö U, et al. Methodologi-
Clinical validation of a new subtraction radiography tech- cal aspects and quantitative adjuncts to computerized sub-
nique for periodontal bone loss detection. J Periodontol traction radiography. J Periodontal Res 1987;22:125-132.
2000;71:598-605.
10. Cury PR, Sallum EA, Nociti FH Jr, Sallum AW, Jeffcoat Correspondence: Dr. Patricia R. Cury, R. Abolição 1827,
MK. Long-term results of guided tissue regeneration ther- Campinas, SP, CEP: 13045-610, Brazil. Fax: 55-19-3237-3611;
apy in the treatment of Class II furcation defects: A ran- e-mail: curyp@usp.br.
domized clinical trial. J Periodontol 2003;74:3-9.
11. Jeffcoat MK, Reddy MS, van den Berg HR, Bertens E. Accepted for publication December 25, 2003.
Quantitative digital subtraction radiography for the
assessment of peri-implant bone change. Clin Oral
Implants Res 1992;3:22-27.
12. Byrd V, Mayfield-Donahoo T, Reddy MS, Jeffcoat MK.
Semiautomated image registration for digital subtrac-
tion radiography. Oral Surg Oral Med Oral Pathol Oral
Radiol Endod 1998;85:473-478.
13. Jeffcoat MK, Reddy MS, Webber RL, Williams RC,
Rüttimann UE. Extraoral control of geometry for digi-
tal subtraction radiography. J Periodontal Res 1987;22:
396-402.
14. Hausmann E, Christersson L, Dunford R, Wikesjö U,
Phyo J, Genco RJ. Usefulness of subtraction radiogra-
phy in the evaluation of periodontal therapy. J Periodontol
1985;56:4-7.
15. Wenzel A, Warrer K, Karring T. Digital subtraction radi-
ography in assessing bone changes in periodontal
defects following guided tissue regeneration. J Clin Peri-
odontol 1992;19:208-213.

1149