Original Paper
Caries Res 2014;48:566–574
DOI: 10.1159/000357596
A Comparative Study of Different
Radiographic Methods for Detecting
Occlusal Caries Lesions
Elif Tarım Ertas a Ebru Küçükyılmaz b Hüseyin Ertaş c Selçuk Savaş b
Meral Yırcalı Atıcı a   
Departments of a Oral and Maxillofacial Radiology, b Pedodontics and c Endodontics, Faculty of Dentistry,
Izmir Katip Çelebi University, Izmir, Turkey
Key Words
Computed tomography · Dental caries · Diagnosis ·
Digital · Radiography
Abstract
Objectives: The aim of this in vitro study was to compare the
diagnostic accuracy of different radiographic imaging mo-
dalities in detecting occlusal caries lesions. Materials and
Methods: Under standardized conditions, 125 extracted hu-
man permanent molar teeth with sound or occlusal caries
lesions were radiographed using a conventional film system
(F-speed), a direct digital imaging system (complementary
metal oxide semiconductor sensor), an indirect digital imag-
ing system (photostimulable phosphor plate) and a cone
beam computed tomography system (CBCT). Two observers
scored the resultant images for the presence or absence of
caries. Then, the teeth were histologically prepared and a
definite diagnosis was determined by stereomicroscopic as-
sessment. The area under the receiver operating character-
istic curve (Az), sensitivity, specificity and accuracy of each
imaging modality were calculated, as well as the intra- and
interexaminer reproducibility. Results: For both thresholds,
interexaminer agreement were higher for CBCT. For intraex-
aminer agreement, observers had different scores for both
thresholds, but the scores were generally higher for CBCT.
Similar Az values were achieved with all imaging methods at
© 2014 S. Karger AG, Basel
0008–6568/14/0486–0566$39.50/0
E-Mail karger@karger.com
www.karger.com/cre
Received: August 5, 2013
Accepted: November 26, 2013
Published online: July 29, 2014
a diagnostic D1 threshold. The Az values of the CBCT system
were found to be statistically higher than those of the other
imaging modalities at a diagnostic D3 threshold (p > 0.05);
no significant differences were found among the other im-
aging modalities. All radiographic methods showed similar
sensitivities, specificities and accuracy in detecting D1
threshold. The CBCT system showed higher sensitivity and
accuracy in detecting dentine lesions. Conclusions: Within
the limitations of this study, CBCT exhibited better perfor-
mance in detecting deep occlusal caries lesions than the
other radiographic systems. © 2014 S. Karger AG, Basel
In recent years, the prevalence of dental caries has de-
clined worldwide in a remarkable way [Rathore et al.,
2012]. However, accurately diagnosing occlusal pit and
fissure caries and detecting incipient lesions are important
issues for clinicians [Souza-Zaroni et al., 2006]. Using a
visual/tactile method when diagnosing occlusal caries is
the oldest and most common method preferred by den-
tists in clinical practice [Rathore et al., 2012]. Since occlu-
sal caries can progress without visible breakdown of the
enamel structure [Poorterman et al., 2000], visual exami-
nation alone is not always sufficient for diagnosing occlu-
sal caries. In the literature, although visual inspection has
been reported to be highly specific [Heinrich-Weltzien et
169.230.243.252 - 11/27/2014 3:27:14 PM
Elif Tarım Ertaş, DDS, PhD
UCSF Library & CKM
Department of Oral and Maxillofacial Radiology
Faculty of Dentistry, Izmir Katip Çelebi University
Downloaded by:
TR–35180 Izmir (Turkey)
E-Mail dteliftarim @ yahoo.com
al., 2005; Reis et al., 2006], these methods are subjective,
which normally leads to low sensitivity and reliability re-
sults [Ketley and Holt, 1993; Costa et al., 2002].
In order to overcome difficulties during diagnosis and
enable better detection of occlusal caries, radiographic
examinations, especially conventional film and digital in-
traoral radiography, are the most easily accessible tech-
niques for improving caries detection in routine clinical
practice [Kamburoğlu et al., 2010, 2011]. Recent develop-
ments in imaging systems and the production of new sen-
sor types and advanced software programs offer increas-
ing clinical advantages [Kayipmaz et al., 2011]. Although
the image resolution of conventional film is superior to
that of digital images [Parks and Williamson, 2002], this
technique requires more radiation to produce an image
of diagnostic quality; therefore, many professionals are
now replacing conventional film radiographs with digital
radiography, due to its many advantages [Cederberg et
al., 1998; Paurazas et al., 2000; Williams, 2001; Kitagawa
et al., 2003; Naoum et al., 2003; Kayipmaz et al., 2011].
There are several digital radiographic systems currently
used in dental practice as an alternative to film-based radi-
ography [Pontual et al., 2010]. The most common direct
digital imaging systems use solid state sensors – either a
charge-coupled device (CCD) or a complementary metal
oxide semiconductor (CMOS); indirect digital imaging
systems use photostimulable phosphor (PSP) plates, also
known as storage phosphor plates [Kamburoğlu et al., 2010].
Although digital systems have a number of advantages
[Wenzel, 1995; Hintze et al., 2002; Jacobsen et al., 2004],
such as lower exposure dose, reduced working time from
image exposure to image display (no wet processing is
involved), lack of destroyed processing artifacts often ex-
perienced with conventional film, and possible image
quality enhancements, such as contrast and density mod-
ulation, which might increase diagnostic accuracy [Wen-
zel, 1995, 2000; Hintze et al., 2002; Pai and Zimmerman,
2002], all of these systems are able to provide two-dimen-
sional (2D) information about dental tissues and diseases
[Wenzel, 2000]. Another disadvantage of 2D radiographs
is that the apparent depth can also vary as a function of
X-ray beam angulation [van der Stelt et al., 1989; Chad-
wick et al., 1999] and radiographic density [Versteeg et
al., 1997], which can lead to variations in perception. The
perceived lesion depth can lead dentists to erroneously
believe that caries has progressed or regressed, resulting
in unnecessary restorative intervention or delay in treat-
ment [Akdeniz and Gröndahl, 2005].
Due to the high demand for a technique that can pro-
vide three-dimensional (3D) data at the tooth level
Different Radiographic Methods for
Detecting Caries
[Kamburoğlu et al., 2011], at a lower cost and with lower
absorbed doses than with the conventional computed to-
mography used in medical radiology [Tyndall and Ra-
thore, 2008], a new technology that uses a 2D sensor and a
cone-shaped beam in place of the fan-shaped X-ray beam
used for conventional computed tomography has been de-
veloped [Haiter-Neto et al., 2008]. The development of
cone beam computed tomography (CBCT) has been revo-
lutionary in that this technology [Rathore et al., 2012] of-
fers a number of potential advantages over conventional
tomography, including easier image acquisition, higher
image accuracy, fewer artifacts, lower effective radiation
doses (up to 15 times lower than those of conventional
computed tomography scans), faster scan times and great-
er cost effectiveness [Scarfe and Farman, 2008; Tyndall and
Rathore, 2008]. This technique could be applied in several
dental diagnostic areas, such as implant treatment, cranio-
facial anomalies, endodontic treatment, orthodontics,
periodontology [Arai et al., 1999; Ziegler et al., 2002; Dan-
forth, 2003] and caries diagnosis [Kayipmaz et al., 2011].
In the literature, the results of studies using CBCT for car-
ies detection are not consistent with each other. In some
studies, promising results have been reported in the detec-
tion of caries lesions [Haiter-Neto et al., 2008], while in
another study, a limited CBCT system was found to be
superior to conventional film and storage phosphor radi-
ography for the in vitro assessment of approximate caries
lesion depth [Akdeniz et al., 2006]. Tsuchida et al. [2007]
and Haiter-Neto et al. [2008] had raters score images of
teeth made with a CBCT unit and found no benefit over
film for detecting incipient proximal surface caries [Young
et al., 2009]. Thus, there are no clear conclusions regarding
the value of CBCT for detecting caries in the literature. The
purpose of the present study was to compare the caries di-
agnostic accuracy of conventional film radiography (F-
speed film), a direct digital imaging system (CMOS sen-
sor), an indirect digital imaging system (PSP plate) and a
CBCT unit that provides high-resolution images with the
smallest voxel size (0.075 mm) on the market for the in vi-
tro determination of occlusal caries.
Materials and Methods
This study was approved by the Research Ethics Committee of
Izmir Katip Çelebi University (registration number 2013-39).
Sample Selection
A total of 125 extracted permanent molar teeth exhibiting com-
plete root formation were included in this study. Altered physical
properties in the tooth structure, large cavitated surfaces and den-
169.230.243.252 - 11/27/2014 3:27:14 PM
Caries Res 2014;48:566–574 567
UCSF Library & CKM
DOI: 10.1159/000357596
Downloaded by:
tal restorations were not included in the study sample. The occlu-
sal surfaces ranged from sound to varying degrees of fissure, dis-
coloration and possible microscopic breakdown of the surface
structure; however, none of the teeth showed macroscopic signs of
cavity formation with exposure into the dentine. The teeth were
stored in 0.1% thymol solution for less than 3 months from the
time of extraction. For the study, the teeth were radiographed us-
ing four different radiographic methods.
Conventional Film Radiography
A bitewing technique using size 2 (3 × 4 mm) F-speed film (CF-
SPEEDX, Medex Medical Imaging, Nice, France) and standard-
ized bitewing projection geometry was used. The intraoral X-ray
unit (eXTtend; MyRay, Imola, Italy) was operated at 65 kV, 7 mA
with 2.5 mm aluminium equivalent filtration. Focus-to-film dis-
tance was 30 cm. Each tooth was mounted in a block of silicone
paste to ensure a reproducible geometry. While the vinyl polysi-
loxane putty was still soft, the film holder was pressed into it, and
once hardened, the putty allowed quick realignment of the speci-
men as well as CMOS sensor, PSP plate and F-speed conventional
films. A 20-mm-thick soft tissue equivalent Plexiglas block was
placed close to the tooth and facing the X-ray tube to simulate scat-
ter radiation and beam attenuation from soft tissues. The F-speed
films were exposed for 0.25 s to generate an optimal density sub-
jectively for caries detection [Sogur et al., 2011]. The exposed films
were automatically processed (Velopex Extra-X; Medivance In-
struments Ltd., London, UK) using fresh Kodak developer and
fixer solutions.
Direct Digital Radiography – CMOS Sensor
Standardized images of the teeth were obtained using the same
intraoral X-ray unit, but with a size 1 (37 × 24 mm) CMOS sensor
(DIGORA Toto; SOREDEX, Milwaukee, Wisc., USA) with the
same standardized projection geometry. The CMOS sensors were
exposed for 0.12 s to generate an optimal density subjectively for
caries detection [Sogur et al., 2011]. The direct digital imaging sys-
tem provides three types of diagnostic modes (dentoenamel, peri-
odontal and endodontic) at two levels (high and low) of spatial
resolution. In this study, the exposures with the CMOS sensor were
performed in the dentoenamel high-resolution mode, which was
recommended by the manufacturer to provide high-contrast im-
ages for caries detection.
Indirect Digital Radiography – PSP Plate System
Standardized images of the teeth were obtained using the same
intraoral X-ray unit, but with size 2 (31 × 41 mm) VistaScan blue
storage phosphor plates (Dürr Dental, Bietigheim-Bissingen,
Germany) with the same standardized projection geometry. The
PSP plates were exposed for 0.12 s to generate an optimal density
subjectively for caries detection [Sogur et al., 2011]. The plates
were later scanned in a VistaScan scanner. The PSP plates were
stored in lightproof envelopes during the exposure and scanned
immediately after exposure using the VistaScan scanner. The high-
resolution scan mode was selected from the scanner setup menu,
as recommended by the manufacturer for most diagnostic tasks.
CBCT
Each tooth block was also radiographed using a NewTom 5G
CBCT system (Verona, Italy) with a 6 × 6 cm field of view in the
high-resolution denture scan mode. The voxel size was 0.075 mm3
568 Caries Res 2014;48:566–574
DOI: 10.1159/000357596
at a fixed 110 kVp setting, automated adjusted milliamperes and a
scan time of 36 s. The volumetric data from the CBCT system were
reconstructed and sectioned into 0.075-mm pieces in the mesio-
distal tooth plane.
Evaluation of Radiographic Methods
The images were organized into groups of images from a par-
ticular exposure setting and system. Two observers (one pedodon-
tist and one oral and maxillofacial radiologist) independently
viewed the image groups in random order. The oral and maxillofa-
cial radiologist with 10 years of experience in image interpretation
provided a training session to familiarize the other observer with
the presentation of all imaging methods and was given brief infor-
mation about the use of the NNT viewing software (version 3.10,
QR srl, Verona, Italy) on how to observe the CBCT images. Then
the observers discussed and scored a number of teeth with carious
lesions with different radiographic methods for calibration. The
digital and CBCT images were displayed using the dedicated soft-
ware of each imaging system incorporated into the same computer.
Observation conditions were optimized by using the same com-
puter monitor to display the images; the display ratio was 1:1. View-
ing distance was kept constant, at about 50 cm, for all observers, and
the lights were dimmed during the observations. The observers
were not given the option of performing any image enhancements
to avoid the production of a variety of different digital images.
Conventional film radiographs were examined on a light box
and at 2× magnification. A period of at least 1 day separated each
viewing session. All tooth surfaces were examined for the presence
of carious lesions on the occlusal surfaces, using a five-point con-
fidence rating scale: 0 = no caries; 1 = radiolucency extending to
the outer half of the enamel; 2 = radiolucency extending to the in-
ner half of the enamel; 3 = radiolucency extending to the outer half
of the dentine; 4 = radiolucency extending to the inner half of the
dentine. Intraobserver agreement was assessed by having each ob-
server view all images twice, with a 2-week interval between view-
ing to eliminate memory bias.
Histological Validation
The teeth were individually embedded in acrylic (Vipcril; Vipi,
São Paulo, Brazil) and serially sectioned into 700-μm-thick sections
in the mesiodistal direction, using a water-cooled 200-μm diamond
band. The histological examination was performed by one of the
study authors (E.K.), with experience from several previous in vitro
studies, using a stereomicroscope with a magnification of 10–20×
under reflected light (Olympus SZ61, Tokyo, Japan).
Caries was defined as being present when demineralization was
observed, seen as a white or discolored (yellow/brown) area. The
histological criteria for caries lesion depth were: 0 = no caries; 1 =
demineralization extending to the outer half of the enamel; 2  =
demineralization extending to the inner half of the enamel; 3  =
demineralization extending to the outer half of the dentine; 4 =
demineralization extending to the inner half of the dentine. As a
measure of histological assessment reliability, all of the sections
were re-examined after 10 days. Almost perfect agreement was
reached between the first and second histological examinations.
Statistical Analysis
The analyses of the caries detected on the radiographs were
performed at two different thresholds: enamel and dentine caries
lesions (D1 threshold; sound vs. decayed) and dentine caries le-
169.230.243.252 - 11/27/2014 3:27:14 PM
Tarım Ertas/Küçükyılmaz/Ertaş/Savaş/
UCSF Library & CKM
Yırcalı Atıcı
Downloaded by:
 Color version available online
among each other, receiver operating characteristic (ROC) analy-
ses were performed, and the area under the ROC curve (Az) at the
1.0 D1 and D3 thresholds was calculated. Comparison of ROC curves
was calculated with the statistics program MedCalc 9.3.0.0 (Mar-
iakerke, Belgium). The analyses were performed using the SPSS
0.8 statistics program for Windows, version 11.5 (SPSS Inc., Chicago,
Ill., USA). For all statistical analyses, the level of significance was
p < 0.05.
0.6
Sensitivity

Results
0.4
Distribution of Lesions
Source of the curve
Conventional film A total of 125 occlusal surfaces were examined in the
0.2 PSP study. According to the histological examination, the sta-
CMOS
CBCT tus of the 125 occlusal surfaces was: 31 (24.8%) sound, 11
Reference line (8.8%) with caries lesions extending into the outer half of
0 the enamel, 47 (37.6%) with caries lesions extending into
0 0.2 0.4 0.6 0.8 1.0
1 – Specificity
the inner half of the enamel, 29 (23.2%) with caries lesions
extending into the outer half of the dentine, and 7 (5.6%)
with caries lesions extending into the inner half of the
dentine.
Fig. 1. Area under the ROC curve values of all radiographic detec-
tion methods at the D1 threshold. Diagonal segments are pro-
duced by ties. Occlusal Lesions at the D1 Threshold
The sensitivity scores of the CBCT images of occlusal
lesions at the D1 threshold were higher than those of the
Table 1. Sensitivity, specificity, accuracy and Az scores of all radio-
other imaging methods, while the specificity scores were
graphic detection methods at the D1 threshold lowest. However, the diagnostic accuracy of all radio-
graphic systems was assessed using the Az, and similar
Test method Az Sensitivity Specificity Accuracy accuracy scores were obtained with all imaging methods
at the D1 threshold. A comparison of the Az values showed
Conventional film 0.851 0.798 0.903 0.824 that the differences between the radiographic methods
PSP 0.824 0.713 0.935 0.768
CMOS 0.834 0.830 0.839 0.832 were not significant (p = 0.401 between conventional film
CBCT 0.801 0.926 0.677 0.864 and PSP plate, p = 0.725 between conventional film and
CMOS sensor, p = 0.338 between conventional film and
CBCT, p = 0.840 between PSP plate and CMOS sensor,
p  = 0.663 between PSP plate and CBCT, p  = 0.447 be-
tween CMOS sensor and CBCT) (fig. 1). Table 1 shows
sions (D3 threshold), based on the histological evaluation as the
gold standard. The appropriate cutoff point for the D1 threshold the sensitivity, specificity, accuracy and Az values of all
was a score of 1 or above, considering the gold standard scores 1, radiographic methods at the D1 threshold.
2 and 3 as evidence of disease for all radiographic methods. The
appropriate cutoff point for the D3 threshold was a score of 3 or Occlusal Lesions at the D3 Threshold
above, considering the gold standard scores 3 and above as evi- The sensitivity scores of the CBCT images of occlusal
dence of disease for all radiographic methods. Data were presented
and analyzed separately for each examiner. lesions at the D3 threshold were higher than those of the
Inter- and intraexaminer reliabilities were calculated using Co- other imaging methods. At the D3 threshold, similar
hen’s kappa test after collapsing the results into two categories: D1 specificity and accuracy scores were obtained with all im-
and D3 thresholds. For each observer and each radiographic mo- aging methods. A comparison of the Az values showed
dality, the sensitivity (cumulative percentage of carious enamel le- significant differences between CBCT and the other ra-
sions identified among those that had carious lesions), specificity
(cumulative percentage of sound surfaces identified among those diographic methods (p  = 0.029 between conventional
who had sound surfaces) and accuracy (percentage of correct film and CBCT, p = 0.023 between PSP plate and CBCT,
scores) were computed. To compare the performance of methods p = 0.014 between CMOS sensor and CBCT). There were
169.230.243.252 - 11/27/2014 3:27:14 PM

Different Radiographic Methods for Caries Res 2014;48:566–574 569

UCSF Library & CKM

Detecting Caries DOI: 10.1159/000357596

Downloaded by:
 Color version available online
Table 2. Sensitivity, specificity, accuracy and Az scores of all radio-
1.0 graphic detection methods at the D3 threshold

Test method Az Sensitivity Specificity Accuracy

0.8
Conventional film 0.739a 0.500 0.978 0.840
PSP 0.730a 0.472 0.989 0.840
0.6
CMOS 0.717a 0.444 0.989 0.832
0.875b
Sensitivity

CBCT 0.806 0.944 0.904

Different superscript letters show a significant difference be-

0.4
tween methods.
Source of the curve
Conventional film
0.2 PSP
CMOS
CBCT Table 3. Interexaminer kappa scores (with standard errors) of all
Reference line radiographic detection methods at the D1 and D3 thresholds
0
0 0.2 0.4 0.6 0.8 1.0
Test method D1 D3
1 – Specificity

Conventional film 0.473 (0.071) 0.822 (0.070)

PSP 0.654 (0.068) 0.804 (0.071)
Fig. 2. Area under the ROC curve values of all radiographic detec- CMOS 0.518 (0.083) 0.763 (0.079)
tion methods at the D3 threshold. Diagonal segments are pro- CBCT 0.833 (0.061) 0.980 (0.020)
duced by ties.

no differences among the other methods at the D3 thresh- that intraoral analogue X-ray film, direct and indirect
old (p = 0.899 between conventional film and PSP plate, digital X-ray modalities (CMOS sensor and PSP plate)
p = 0.732 between conventional film and CMOS sensor, and CBCT performed similarly in detecting occlusal car-
p = 0.560 between PSP plate and CMOS sensor) (fig. 2). ies at the D1 threshold, and that CBCT performed better
Table 2 shows the sensitivity, specificity, accuracy and Az at the D3 threshold.
values of all radiographic methods at the D3 threshold. In the present study, the occlusal surfaces ranged
In the kappa analysis of interexaminer agreement be- from sound to varying degrees of fissure, discoloration
tween the observers, conventional film radiography had and possible microscopic breakdown of the surface
the lowest scores at the D1 threshold, while the CBCT im- structure; however, none of the teeth showed macro-
ages had higher scores at both the D1 and D3 thresholds scopic signs of cavity formation with exposure into den-
compared with the other methods. At the D3 threshold, tine, as we believe that if diagnostic differences be-
repeatability between the observers was similar for all im- tween radiographic systems are to be found, their accu-
aging methods and higher than the scores obtained at the racy in detecting subtle pathological changes should be
D1 threshold (table 3). tested.
There was a high level of intraexaminer agreement be- An in vitro model was preferred in this radiological
tween the two assessments of the observers for the CBCT study, as ideal patient positioning is not always possible
images at the D3 threshold. The overall intraexaminer in an in vivo study and absolute reproducibility is limited.
scores were compatible between the observers at both In addition, image quality may vary from one patient to
thresholds. Table 4 shows the intraexaminer agreement another. Another advantage of this in vitro model is that
scores of both observers at both thresholds. the occlusal surfaces can be exposed to X-rays repeatedly
and ideal positioning of the specimen with exact repro-
ducibility is possible. However, the results of the present
Discussion study might not correlate with clinical situations. This is
because in clinical settings, movement of the patient de-
The present research focused mainly on the diagnostic creases image resolution, restorations in the occlusal
accuracy and reliability of observers regarding images ob- plane can cause metallic streaking artifacts in the occlusal
tained using different X-ray systems. This study found plane, and the other head and neck structures can result
169.230.243.252 - 11/27/2014 3:27:14 PM

570 Caries Res 2014;48:566–574 Tarım Ertas/Küçükyılmaz/Ertaş/Savaş/

UCSF Library & CKM

DOI: 10.1159/000357596 Yırcalı Atıcı

Downloaded by:
 Table 4. Intraexaminer kappa scores (with standard errors) of all radiographic detection methods at the D1 and
D3 thresholds
Test method D1
examiner 1
Conventional film 0.872 (0.044)
PSP 0.824 (0.051)
CMOS 0.513 (0.077)
CBCT 0.776 (0.067)
in scattering into the field of interest, thereby reducing
contrast [Ricketts et al., 1995].
The accuracy of 2D systems is well established in the
literature [Rathore et al., 2012]; however, the accuracy of
both intraoral analogue X-ray film and digital X-ray sen-
sor measurements is limited by the 2D nature of the tech-
nology. The lower radiation doses, superimposition of
anatomical structures and patient-related factors that af-
fect caries diagnosis are inevitable factors with 2D sys-
tems [Kamburoğlu et al., 2011]. In addition, the differ-
ences in mass between small, incipient lesions and the
surrounding tissues are so small that they do not reflect
density differences with 2D images [Akdeniz and Grön-
dahl, 2005]. In the past decade, the use of CBCT in den-
tistry has become more and more widespread, as the
CBCT technology overcomes the irradiation geometry
problems that can cause errors in caries diagnosis with 2D
imaging [Akdeniz and Gröndahl, 2005].
In this study, the observers were calibrated for each
radiographic interpretation, but some differences be-
tween the observers still occurred, which may be due to
their levels of experience and training with these radio-
graphic methods.
In the present study, ROC analysis was used to evalu-
ate the diagnostic performance of four radiographic
methods. In this analysis, significant differences among
the areas under the ROC curves of the competing tech-
niques were compared [Kantor et al., 1989], and it was
found that the area under the curve (AUC) reflected di-
agnostic performance more comprehensively than sensi-
tivity and specificity did, which are considered only one
cutoff point [Kositbowornchai et al., 2004]. The ROC
curve distinguishes between the inherent capacities of the
observers to under- and overread when interpreting im-
aging; therefore, this analysis provides the most mean-
ingful approach to comparing the diagnostic perfor-
mance of two or more different imaging modalities [Ra-
thore et al., 2012]. In this analysis, an area of 1 represents
Different Radiographic Methods for
Detecting Caries
D3
examiner 2 examiner 1 examiner 2
0.610 (0.073) 0.546 (0.111) 0.647 (0.100)
0.727 (0.059) 0.548 (0.106) 0.834 (0.072)
0.714 (0.064) 0.666 (0.093) 0.965 (0.035)
0.935 (0.037) 0.918 (0.040) 0.938 (0.035)
a perfect test, and anything near 0.5 is a poor test result
[Hintze et al., 2003]. In the present study, the AUC values
of the four radiographic methods were around 0.80–0.85
at the D1 threshold, indicating that none of the methods
was superior to the others and that they performed well
at the D1 threshold. In addition, the CBCT AUC values
were higher than those of the other methods at the D3
threshold, and the scores were close to those of the other
methods.
In recent years, several studies have been carried out
to evaluate the accuracy of CBCT in detecting caries le-
sions on proximal and occlusal surfaces, with varying re-
sults [Haiter-Neto et al., 2008; Young et al., 2009;
Kamburoğlu et al., 2010; Qu et al., 2011]. In 2007, Kalath-
ingal et al. published a study comparing a CBCT device
(SIDEXIS sensor; Sirona Dental Systems, Bensheim,
Germany) and conventional film radiography in the de-
tection of proximal caries, and their results showed no
differences between the two methods. In the same year,
Tsuchida et al. reported that the accuracy of the 3D Ac-
cuitomo in evaluating incipient proximal caries was not
superior to that of intraoral films. Similar to their previ-
ous finding, Haiter-Neto et al. [2008] reported no differ-
ences in specificity or overall true scores among the meth-
ods when comparing the diagnostic accuracy of two
CBCT systems – NewTom 3G (Verona, Italy) and 3DX
Accuitomo – with one digital PSP (DIGORA fmx) and
one conventional film system (Kodak Insight) in detect-
ing occlusal caries.
Contrary to previous findings, Young et al. [2009] re-
ported that Accuitomo 3DX images were superior to
CCD projection images in detecting lesions extending
into the dentine on occlusal and proximal surfaces. How-
ever, the authors concluded that the 3DX images did not
offer significantly superior information for detecting
proximal surface caries limited to the enamel compared
with CCD projection imaging. The possible explanation
for these differences in study results is that both Tsuchida
169.230.243.252 - 11/27/2014 3:27:14 PM
Caries Res 2014;48:566–574 571
UCSF Library & CKM
DOI: 10.1159/000357596
Downloaded by:
et al. [2007] and Haiter-Neto et al. [2008] used a popula-
tion of teeth in which the vast majority of the proximal
surface lesions were limited to the enamel, whereas Young
et al. [2009] evaluated both enamel and dentinal lesions
equally. The different distribution of lesion depths would
make detecting lesions more challenging, especially when
they are located in enamel. In a recent study by Rathore
et al. [2012], no statistically significant differences were
reported between the Sirona Galileos (Sirona Dental Sys-
tems) CBCT and conventional film radiography in de-
tecting occlusal caries.
The results of the present study are consistent with the
data in the literature in that there were no differences be-
tween the four imaging methods in detecting occlusal
caries at the D1 threshold (sound/decayed), whereas
CBCT was superior in detecting dentine lesions.
CBCT systems offer a feature whereby an area can be
evaluated in three planes at the same time (axial, sagittal
and coronal), and sections can be created with different
slice thicknesses and intervals. In the present study, we
chose the mesiodistal plane to validate occlusal surfaces,
as in this section, occlusal surface demineralization can
be evaluated clearly [Haiter-Neto et al., 2008]. However,
it is critical that if the operator fails to choose the slice
that shows the deepest extension of the lesion, he/she
may misdiagnose the lesions [Akdeniz and Gröndahl,
2005]. It has been suggested that the small voxel sizes
and thin slices are the potential advantages of a CBCT
system in caries diagnosis [Kayipmaz et al., 2011]. The
CBCT device used in the present study (NewTom 5G)
offers the smallest voxel size (0.075 mm3) available on
the market, and this is the first study in the literature to
evaluate accuracy at the smallest voxel size. However,
no statistically significant differences were found among
the four radiographic methods in the diagnosis of oc-
clusal caries lesions at the D1 threshold, even with the
smallest voxel size. Consistent with this finding, in a re-
cent study, Kamburoğlu et al. [2010] compared intra-
oral digital CCD sensor images and CBCT images ob-
tained at different voxel resolutions in the in vitro detec-
tion of occlusal caries, and the authors concluded that
there were no statistically significant differences be-
tween ultra-resolution images and high- and low-reso-
lution images.
In the present study, although the CBCT images had
the highest sensitivity scores, the CBCT specificity scores
were lowest at the D1 threshold, which means that CBCT
performs better when detecting true carious lesions, but
weakly when detecting real sound surfaces. The present
finding is consistent with the result of Haiter-Neto et al.
572 Caries Res 2014;48:566–574
DOI: 10.1159/000357596
[2008], who reported that the sensitivity of the two CBCT
systems (Accuitomo and NewTom) was high, but that
their specificity was low, resulting in an increase in false-
positive results. Similarly, in their study, Young et al.
[2009] found more false-positive results in intact teeth,
thus resulting in significantly lower average specificity
scores for the 3DX images than for the CCD images. A
possible explanation for this might be the beam-harden-
ing artifacts that appear in the pericoronal area and may
mislead observers in distinguishing pathologies with low
density, such as caries lesions [Tsuchida et al., 2007]. In
addition, sound dentine under some cusps may appear
artifactually more radiolucent than the surrounding den-
tine on the CBCT images, and this difference between
enamel and dentine density might lead observers to mis-
interpret these areas as dentinal occlusal caries, thereby
elevating the sensitivity scores while depressing the spec-
ificity scores [Young et al., 2009; Kayipmaz et al., 2011].
It is possible that better training of observers regarding
this misinterpretation might improve the observers’ spec-
ificity scores without significantly depressing their sensi-
tivity scores [Young et al., 2009].
In clinical practice, when referring a patient for CBCT
examination, one of the most important issues to keep in
mind is the effective radiation dose. Although the effec-
tive dose is somewhat lower than that of medical com-
puted tomography, the dose is relatively higher than in
conventional imaging alternatives and intraoral exami-
nations. It is critical that the potential patient benefits
from a radiographic examination be balanced against the
risk of exposure to ionizing radiation [Ludlow et al.,
2003], and it is fundamental that diagnostic radiology and
CBCT procedures should be reserved for selected cases
[Farman, 2005]. It should also be noted that CBCT is not
considered suitable for routine caries diagnosis, because
the large majority of patients have metallic restorations
that cause artifacts due to beam hardening and scatter
that simulates recurrent caries, and because the quality of
CBCT images can be seriously affected by patient motion
[Clifton et al., 1998].
In conclusion, the NewTom 5G CBCT exhibited high-
er sensitivity for detecting occlusal lesions than the intra-
oral systems did, but the overall accuracy scores were sim-
ilar at the D1 threshold. CBCT was more accurate in de-
tecting lesions at the D3 threshold than the intraoral
systems. However, it can be concluded that because of the
possible disadvantages of CBCT systems in clinical prac-
tice, including relatively higher effective radiation doses
as well as possible misdiagnosis due to the artifacts caused
by movement and metallic artifacts, intraoral radiogra-
169.230.243.252 - 11/27/2014 3:27:14 PM
Tarım Ertas/Küçükyılmaz/Ertaş/Savaş/
UCSF Library & CKM
Yırcalı Atıcı
Downloaded by:
phy methods are found to be sufficient and have practical,
economical and low-dose features that are valuable for
the routine assessment of caries.
Author Contributions
Conception and design of study: E. Tarım Ertas. Acquisition
of data: S. Savaş, M. Yırcalı Atıcı. Analysis and/or interpretation
of data: E. Tarım Ertas, E. Küçükyılmaz, H. Ertaş. Drafting of
References
Akdeniz BG, Gröndahl HG: Proximal caries
depth: agreement between limited cone-beam
CT, storage phosphore and film radiography.
Hacettepe Dishekimligi Fakultesi Dergisi
2005;29:7–12.
Akdeniz BG, Gröndahl HG, Magnusson B: Accu-
racy of proximal caries depth measurements:
comparison between limited cone beam com-
puted tomography, storage phosphor and
film radiography. Caries Res 2006; 40: 202–
207.
Arai Y, Tammisalo E, Iwai K, Hashimoto K, Shi-
noda K: Development of a compact computed
tomographic apparatus for dental use. Dento-
maxillofac Radiol 1999;28:245–248.
Cederberg RA, Tidwell E, Frederiksen NL, Ben-
son BW: Endodontic working length assess-
ment. Comparison of storage phosphor digi-
tal imaging and radiographic film. Oral Surg
Oral Med Oral Pathol Oral Radiol Endod
1998;85:325–328.
Chadwick BL, Dummer PM, van der Stelt PF:
The effect of alterations in horizontal X-ray
beam angulation and bucco-lingual cavity
width on the radiographic depth of approxi-
mal cavities. J Oral Rehabil 1999; 26: 292–
301.
Clifton TL, Tyndall DA, Ludlow JB: Extraoral ra-
diographic imaging of primary caries. Dento-
maxillofac Radiol 1998;27:193–198.
Costa AM, Yamaguti PM, De Paula LM, Bezerra
AC: In vitro study of laser diode 655 nm diag-
nosis of occlusal caries. ASDC J Dent Child
2002;69:249–253, 233.
Danforth RA: Cone beam volume tomography: a
new digital imaging option for dentistry. J Ca-
lif Dent Assoc 2003;31:814–815.
Farman AG: ALARA still applies. Oral Surg Oral
Med Oral Pathol Oral Radiol Endod 2005;
100:395–397.
Haiter-Neto F, Wenzel A, Gotfredsen E: Diagnos-
tic accuracy of cone beam computed tomog-
raphy scans compared with intraoral image
modalities for detection of caries lesions.
Dentomaxillofac Radiol 2008;37:18–22.
Heinrich-Weltzien R, Kuhnisch J, Ifland S,
Tranaeus S, Angmar-Mansson B, Stosser L:
Detection of initial caries lesions on smooth
surfaces by quantitative light-induced fluo-
Different Radiographic Methods for
Detecting Caries
the manuscript: E. Tarım Ertas. Critical revision of the manu-
script for important intellectual content: H. Ertaş, E. Tarım
Ertas. Approval of the version of the manuscript to be pub-
lished: E. Tarım Ertas, H. Ertaş, E. Küçükyılmaz, S. Savaş, M.
Yırcalı Atıcı.
Disclosure Statement
The authors declare that they have no conflicts of interest.
rescence and visual examination: an in vivo
comparison. Eur J Oral Sci 2005;113:494–498.
Hintze H, Frydenberg M, Wenzel A: Influence of
number of surfaces and observers on statisti-
cal power in a multiobserver ROC radio-
graphic caries detection study. Caries Res
2003;37:200–205.
Hintze H, Wenzel A, Frydenberg M: Accuracy of
caries detection with four storage phosphor
systems and E-speed radiographs. Dento-
maxillofac Radiol 2002;31:170–175.
Jacobsen JH, Hansen B, Wenzel A, Hintze H: Re-
lationship between histological and radio-
graphic caries lesion depth measured in im-
ages from four digital radiography systems.
Caries Res 2004;38:34–38.
Kalathingal SM, Mol A, Tyndall DA, Caplan DJ:
In vitro assessment of cone beam local com-
puted tomography for proximal caries detec-
tion. Oral Surg Oral Med Oral Pathol Oral
Radiol Endod 2007;104:699–704.
Kamburoğlu K, Kurt H, Kolsuz E, Öztaş B, Tatar
I, Çelik HH: Occlusal caries depth measure-
ments obtained by five different imaging
modalities. J Digit Imaging 2011; 24: 804–
813.
Kamburoğlu K, Murat S, Yüksel SP, Cebeci AR,
Paksoy CS: Occlusal caries detection by using
a cone-beam CT with different voxel resolu-
tions and a digital intraoral sensor. Oral Surg
Oral Med Oral Pathol Oral Radiol Endod
2010;109:e63–e69.
Kantor ML, Zeichner SJ, Valachovic RW, Reiskin
AB: Efficacy of dental radiographic practices:
options for image receptors, examination se-
lection, and patient selection. J Am Dent As-
soc 1989;119:259–268.
Kayipmaz S, Sezgin OS, Saricaoglu ST, Can G: An
in vitro comparison of diagnostic abilities of
conventional radiography, storage phosphor,
and cone beam computed tomography to de-
termine occlusal and approximal caries. Eur J
Radiol 2011;80:478–482.
Ketley CE, Holt RD: Visual and radiographic di-
agnosis of occlusal caries in first permanent
molars and in second primary molars. Br
Dent J 1993;174:364–370.
Kitagawa H, Scheetz JP, Farman AG: Comparison
of complementary metal oxide semiconduc-
Caries Res 2014;48:566–574
DOI: 10.1159/000357596
tor and charge-coupled device intraoral X-ray
detectors using subjective image quality. Den-
tomaxillofac Radiol 2003;32:408–411.
Kositbowornchai S, Basiw M, Promwang Y, Mor-
agorn H, Sooksuntisakoonchai N: Accuracy
of diagnosing occlusal caries using enhanced
digital images. Dentomaxillofac Radiol 2004;
33:236–240.
Ludlow JB, Davies-Ludlow LE, Brooks SL: Do-
simetry of two extraoral direct digital imaging
devices: NewTom cone beam CT and Ortho-
phos Plus DS panoramic unit. Dentomaxillo-
fac Radiol 2003;32:229–234.
Naoum HJ, Chandler NP, Love RM: Convention-
al versus storage phosphor-plate digital im-
ages to visualize the root canal system con-
trasted with a radiopaque medium. J Endod
2003;29:349–352.
Pai SS, Zimmerman JL: Digital radiographic im-
aging in dental practice. Dent Today 2002;21:
56–61.
Parks ET, Williamson GF: Digital radiography: an
overview. J Contemp Dent Pract 2002; 3: 23–
39.
Paurazas SB, Geist JR, Pink FE, Hoen MM, Stei-
man HR: Comparison of diagnostic accuracy
of digital imaging by using CCD and CMOS-
APS sensors with E-speed film in the detec-
tion of periapical bony lesions. Oral Surg Oral
Med Oral Pathol Oral Radiol Endod 2000;89:
356–362.
Pontual AA, de Melo DP, de Almeida SM, Bosco-
lo FN, Haiter Neto F: Comparison of digital
systems and conventional dental film for the
detection of approximal enamel caries. Den-
tomaxillofac Radiol 2010;39:431–436.
Poorterman JH, Weerheijm KL, Groen HJ, Kals-
beek H: Clinical and radiographic judgement
of occlusal caries in adolescents. Eur J Oral Sci
2000;108:93–98.
Qu X, Li G, Zhang Z, Ma X: Detection accuracy of
in vitro approximal caries by cone beam com-
puted tomography images. Eur J Radiol 2011;
79:e24–e27.
Rathore S, Tyndall D, Wright J, Everett E: Ex vivo
comparison of Galileos cone beam CT and in-
traoral radiographs in detecting occlusal car-
ies. Dentomaxillofac Radiol 2012; 41: 489–
493.
169.230.243.252 - 11/27/2014 3:27:14 PM
573
UCSF Library & CKM
Downloaded by:
Reis A, Mendes FM, Angnes V, Angnes G, Grande
RH, Loguercio AD: Performance of methods
of occlusal caries detection in permanent
teeth under clinical and laboratory condi-
tions. J Dent 2006;34:89–96.
Ricketts DN, Kidd EA, Smith BG, Wilson RF:
Clinical and radiographic diagnosis of occlu-
sal caries: a study in vitro. J Oral Rehabil 1995;
22:15–20.
Scarfe WC, Farman AG: What is cone-beam CT
and how does it work? Dent Clin North Am
2008;52:707–730.
Sogur E, Baksı BG, Orhan K, Paksoy SC, Dogan S,
Erdal YS, Mert A: Effect of tube potential and
image receptor on the detection of natural
proximal caries in primary teeth. Clin Oral
Investig 2011;15:901–907.
Souza-Zaroni WC, Ciccone JC, Souza-Gabriel
AE, Ramos RP, Corona SA, Palma-Dibb RG:
Validity and reproducibility of different com-
binations of methods for occlusal caries de-
tection: an in vitro comparison. Caries Res
2006;40:194–201.
574 Caries Res 2014;48:566–574
DOI: 10.1159/000357596
Tsuchida R, Araki K, Okano T: Evaluation of a
limited cone-beam volumetric imaging sys-
tem: comparison with film radiography in de-
tecting incipient proximal caries. Oral Surg
Oral Med Oral Pathol Oral Radiol Endod
2007;104:412–416.
Tyndall DA, Rathore S: Cone-beam CT diagnos-
tic applications: caries, periodontal bone as-
sessment, and endodontic applications. Dent
Clin North Am 2008;52:825–841.
van der Stelt PF, Ruttiman UE, Webber RL,
Heemstra P: In vitro study into the influence
of X-ray beam angulation on the detection of
artificial caries defects on bitewing radio-
graphs. Caries Res 1989;23:334–341.
Versteeg KH, Sanderink GC, Velders XL, van
Ginkel FC, van der Stelt PF: In vivo study of
approximal caries depth on storage phosphor
plate images compared with dental X-ray
film. Oral Surg Oral Med Oral Pathol Oral Ra-
diol Endod 1997;84:210–213.
Wenzel A: Current trends in radiographic caries
imaging. Oral Surg Oral Med Oral Pathol
Oral Radiol Endod 1995;80:527–539.
Wenzel A: Digital imaging for dental caries. Dent
Clin North Am 2000;44:319–338.
Williams CP: Digital radiography sensors: CCD,
CMOS, and PSP. Pract Proced Aesthet Dent
2001;13:395–396.
Young SM, Lee JT, Hodges RJ, Chang TL, Elashoff
DA, White SC: A comparative study of high-
resolution cone beam computed tomography
and charge-coupled device sensors for detect-
ing caries. Dentomaxillofac Radiol 2009; 38:
445–451.
Ziegler CM, Woertche R, Brief J, Hassfeld S: Clin-
ical indications for digital volume tomogra-
phy in oral and maxillofacial surgery. Dento-
maxillofac Radiol 2002;31:126–130.
169.230.243.252 - 11/27/2014 3:27:14 PM
Tarım Ertas/Küçükyılmaz/Ertaş/Savaş/
UCSF Library & CKM
Yırcalı Atıcı
Downloaded by: