The purpose of the present study was to determine
the measurement accuracy and subjective image quality for periodontal disease diagnosis
when using two X-ray tube voltages with a digital photostimulable storage phosphor
sensor.
Material and Methods
A digital photostimulable storage phosphor
(PSP) sensor (Vistascan) and a multipulse X-ray generator (Prostyle Intra) with
two tube voltages were used in this study. The front, premolar and molar region
of two adult human cadaver skulls jaws were imaged using the X-ray tube at 63 kV
and 70 kV, both at 8 mA and decreasing exposure times (160 ms, 120 ms and 80 ms).
A standardized exposure protocol containing waxed occlusal keys and an aiming device
ensured proper and reproducible beam alignment. Three observers assessed the digital
radiographs for 31 selected periodontal bone loss sites. Radiographic measurements
were compared to physical measurements (Standard). Subjective ratings of lamina
dura, crater defect and furcation involvement visibility, contrast perception
and bone quality were also performed.
Results
Multiple regression equation of the variables kV and
exposure time demonstrated no significant difference for the periodontal bone level
measurements (P > 0.05). In 90.3% and 96.7% of the measurements for 70 kV and 63
kV respectively, deviation was within 1 mm. The subjective ratings produced similar
findings in terms of image quality for both tube voltages and the three exposure
times.
Conclusions
The results of the present study revealed that tube
voltages of 63 kV and 70 kV provided similar accuracy and image quality for periodontal
disease diagnosis.
dental digital radiographyradiographic image enhancitalicentradiographic phantomalveolar processperiodontal diseasesfurcation defects.INTRODUCTION
Intraoral radiographs are often a necessary adjunct in the diagnosis of periodontal
diseases regarding extent estimation of alveolar bone loss and visualization of
important structures like periodontal ligament space, lamina dura or trabecular
pattern [1-3]. However, the outcome of this radiographic evaluation
is not only depending on exposure parameters but also on image sensor type and viewing
conditions [4-14]. Accuracy of alveolar bone level measuritalicents
on conventional or digital radiographs has been found to lie within 1 to 2 mm deviation
[15-17], the latter resulting in similar [17-19]
or greater accuracy [20-22]. This discrepancy in literature is
most likely due to the many variables in the radiographic chain, the limited standardization
of previous studies and the continuously improving technology. Only one of these
studies actually investigated a range of exposure times for alveolar bone level
measuritalicents [18]. In addition, no studies could be found investigating
beam energy on periodontal bone measuritalicents. These factors are crucial though since
they directly influence radiographic contrast [23,24]. For dental
caries diagnosis, Svenson and Petersson [25] have ditaliconstrated
no significant difference in diagnostic accuracy between pritalicolars and molars using
conventional films exposed at varying tube voltages, although accuracy increased
for molars at higher kilovoltage (kV). Similarly, the alveolar crest and the associated
accuracy of alveolar bone level measuritalicents may be influenced by different tube
voltages since the thickness of the alveolar crest is variable and often affected
by small changes in mineral bone density. Furthermore, when considering possible
dose savings especially when using digital sensors, laboratory studies have suggested
acceptable image quality at a wide range of kV settings [26-28].
The purpose of the present study was to investigate the possible influence of
two different tube voltages on the measuritalicent accuracy of alveolar bone levels
using a digital photostimulable storage phosphor systitalic (intraoral digital phosphor
plates).
MATERIAL AND METHODS
Two human cadaver skulls containing multiple bone loss sites (including irregular
crater patterns) were obtained with permission from the
Department of Anatomy (Catholic University of Leuven,
Leuven, Belgium) and selected for this study. The first skull was obtained from
an adult cadaver head with soft tissues, both upper and lower cadaver jaw were fixed
in a 10% formalin solution. The second subject used was a dry adult skull covered
with custom-made soft tissue simulation. The latter consisted of melted paraffin
wax, Mix D, having similar attenuation properties to human soft tissues [29],
and which was modelled over the maxilla and mandible. For alveolar bone level assessments,
measuritalicents from the citalicento-enamel junction (CEJ) to the alveolar crest were chosen
for the cadaver jaws, but gutta percha fiducials were glued onto the dry skull's
teeth since dehydration of the CEJ could increase observer errors in identifying
the measuritalicent landmark. Small fragments with a central indentation were thus glued
onto the labial and palatal/lingual surfaces of the teeth so their end- or midpoints
could be used for mesial and distal or central bone level measuritalicents (Figure
1). Physical measuritalicents of the alveolar bone levels around each subject's
teeth (mesial and distal, both oral and buccal) were obtained by two observers (department
of oral imaging) using a digital sliding calliper (Mitutoyo, Hants, UK) at the nearest
0.01 mm accuracy. This was done prior to soft tissue simulation and radiographic
exposure for the dry skull, but after radiographic exposure and flap surgery for
the cadaver jaws. The observer's averaged measuritalicents could then function as the
standard (Standard) for the radiographic evaluations.
(A) Radiographic protocol of the dry adult
skull. The soft tissues were simulated with synthetic material (in white),
bone loss fiducials were introduced made from gutta percha fragments and
rigid occlusal imprints ensured radiographic reproducibility. (B) Maxillary
cadaver jaw with formalin fixed tissues.
Standardized periapical radiographs were made using a photostimulable storage
phosphor (PSP) sensor (Vistascan, Dürr Dental GmbH, Bietigheim-Bissingen, Germany)
and a multipulse X-ray generator (Prostyle Intra, Planmeca Oy, Helsinki, Finland)
at two different kV settings (63 and 70 kV), 8 mA, decreasing exposure times (160
ms, 120 ms and 80 ms) and 30 cm focal-film distance. Reproducible projection geometry
was ensured by utilizing the paralleling technique with aiming devices and bite
blocks (XCP, RINN Corporation, Elgin, IL, USA) individualized with waxed imprints
of the teeth. For each jaw region (front-pritalicolar-molar), the bite blocks were covered
with heated green thermoplastic impression compound (Green Sticks, Kerr Corporation,
Orange, CA, USA) for imprinting of the occlusal patterns. A mechanically interlocking
rectangular (4 cm x 3 cm) collimator (Universal Collimator, RINN Corporation, Illinois,
USA) was mounted onto the tube end. A total of 72 radiographs (3 exposure times
x 2 kV settings x 12 periapical skull regions) were thus obtained from the skull
jaws for this study. Imaging plates were read out using the Vistascan laser scanner
at high resolution (20 lp/mm), erased using the bold illumination unit, and exported
in Tagged Image File Format (TIFF) for randomization and evaluation with italicago advanced,
V.3.5.2. image analysis software (Oral Diagnostic Systitalics, Amsterdam, Netherlands).
An example of the radiographic set-up is given in Figure 2.
Standardized PSP radiographs of the dry skull's
mandibular molar region. The images were obtained using two different tube
voltages, 63 kV and 70 kV, at decreasing exposure times.
Thirty-two periodontal bone loss sites, including infrabony defects and furcation
involvitalicents, were chosen for the radiographic measuritalicents and assessed by three
observers specialized in oral imaging, after three calibration sessions. Alveolar
levels were measured with the software tools to the nearest 0.1 mm, in a room with
reduced ambient light, on three 17-inch LCD monitors having antireflective layers
and same screen resolution (1440 x 900 pixels). Contrast and brightness settings
were set to similar percentages. No image enhancitalicent tools were allowed to adjust
the images. Furthermore, the observers were asked to provide subjective ratings
of following periapical radiographs evaluation parameters: lamina dura
delineation, trabecular
pattern depiction, contrast perception and crater and furcation involvitalicent
visibility, using an ordinal scale from 0 to 3 (1 = bad, 2 = medium, 3 = good, 0
= not possible to properly evaluate the variable).
Statistical analysis
In total, 192 radiographic measuritalicents per observer were compared to the Standard.
A 15% repeat of measuritalicents (n = 32) was done at a 1 week interval. Measuritalicent
consistency between and within observers was determined. The absolute differences
(radiographic measures - physical measures as Standard) from the three observers
were then averaged for further analysis. Multiple regression analysis of the dependent
variable periodontal bone measuritalicent and independent variables kV and exposure
time was carried out at a significance level of 5%. For the subjective measuritalicents,
non parametric statistics were used given the ordinal nature of the data. The variables
kV and exposure time were analyzed by a Friedman ANOVA test.
boldAll statistical analyses were carried out using SPSS
V.13.0. statistical software (SPSS Inc., Chicago, IL, USA) and MedCalc v.9.3.2 (MedCalc
Software bvba, Mariakerke, Belgium).
RESULTS
Measuritalicent accuracy
Inter-observer consistency of the 192 periodontal bone measuritalicents was determined
and the reliability analysis ditaliconstrated an intraclass correlation coefficient
(ICC) of 0.96 (95% confidence interval (CI) with 0.95 and 0.97 as upper and lower
bound respectively). Since high correlation was found, the measuritalicents between
observers could be averaged for further analysis. No intra-observer effect was found
when comparing the 15% repeat of measuritalicents (ICC = 0.96, 0.91 – 0.98 at 95% CI).
Table
1 shows the descriptive statistics of the absolute differences from the Standard
for the different variables. Measuritalicent deviations ranged from 0.00 to 2.00
mm from the Standard. The standard deviation (SD) for all variables was found to
be consistent. The bar chart of the absolute differences in Figure
3 revealed a similar pattern for both kVs and although mean deviations seitaliced
to decrease at rising exposure time, this effect was only minimal since the mean's
range was smaller than 0.1 mm.
Descriptive statistics of the periodontal bone level measurementsa for the variables kV (irrespective exposure time) and exposure time (irrespective kV)
Descriptives
Variables kV
Exposure time
kV 63
kV 70
80 ms
120 ms
160 ms
Sample size
96
96
64
64
64
Minimum
0.00
0.00
0.00
0.00
0.00
Maximum
1.97
2.00
1.80
2.00
1.97
Mean
0.44
0.44
0.47
0.45
0.41
95% CI
0.37 - 0.51
0.37 - 0.52
0.39 - 0.54
0.36 - 0.53
0.33 - 0.50
SD
0.33
0.35
0.32
0.34
0.35
aPeriodontal bone level in mm.
CI = confidence interval of the mean; SD = standard deviation.
Bar chart for the absolute differences (abs
diff in mm, y-axis) of the radiographic measuritalicents from the Standard.
The means and error bars at a 95% confidence interval (CI) are shown for
the different exposure times (ms), clustered by the two kV settings (x-axis).
The multiple
regression equation revealed no significant influence of the independent variables
kV (63 and 70) (P = 0.92)and exposure time (80 ms, 120
ms, 160 ms) (P = 0.38) onthe periodontal bone measuritalicents
at a significance level of 5%. When ignoring exposure time, 90.3% of the 70 kV measuritalicents
and 96.8% of the 63 kV ones were within a clinically acceptable threshold level
of 1 mm deviation (see Figure 4). When reducing the clinical threshold
to 0.5 mm, 66.1% for 70 kV and 66.7% for 63 kV were found to be within this limit.
Although no significant difference was found between the two kV settings, the curve
in Figure 4 increases slightly faster for 63 kV, indicating higher
but insignificant accuracy.
Percentage of measuritalicents (y-axis) falling
within a specific distance from the Standard (X-axis), ignoring exposure
time. At least 90% of measuritalicents were 1 mm deviation. A kV of 63 resulted
in a slightly higher accuracy at this threshold level.
Subjective quality evaluation
The results of the Friedman ANOVA test are presented in Table 2.
The 6 groups compared were the combinations of the two kV settings (63 and 70 kV)
with the three exposure times (80, 120 and 160 ms). No significant differences were
found for the subjective ratings of lamina dura delineation (P = 0.42)
or trabecular pattern visibility (P = 0.13), contrast perception (P = 0.19), crater
(P = 0.95) and furcation (P = 0.16) involvitalicent visibility using the different kV
or exposure time settings. The bar chart from the ordinal scores of the trabecular
pattern depiction variable (see Figure 5) did reveal higher scores
at rising exposure times, although this was found to be insignificant. The same
applies for the kV setting of 63 compared to 70 kV. All other variables had similar
bar charts.
Results of the Friedman ANOVA test for the
ordinal scale ratings, established for the periapical radiographs parameters
visibility evaluation in the six different groupsa
Parameters
Ordinal scale ratings (mean rank) in six groups
N
Chi-Square
Df
Pbvalue
1
2
3
4
5
6
Lamina dura
delineation
3.08
3.58
3.58
3.58
3.58
3.58
6
5.00
5
0.42c
Trabecular pattern depiction
2.33
3.25
3.25
3.75
3.75
4.67
6
8.63
5
0.13c
Contrast perception
3.00
3.00
3.00
3.50
4.50
4.00
6
7.50
5
0.19c
Bone crater involvement
3.33
3.33
3.33
3.33
3.83
3.83
6
1.11
5
0.95c
Furcation defect involvement
2.83
2.33
3.83
3.33
4.33
4.33
6
8.00
5
0.16c
aSix groups are combination of two kV settings and three exposure times: 1 = 70 kV, 80 ms; 2 = 63, kV 80 ms; 3 = 70 kV, 120 ms; 4 = 63 kV, 120 ms; 5 = 70 kV, 160 ms; 6 = 63 kV, 160 ms.
bTested between the groups.
cThe mean differences are not significant at 95% significance level.
Df = degree of freedom.
Bar chart for the subjective quality ratings
of trabecular pattern depiction. Ordinal scores of the observers (y-axis)
are plotted by exposure times (ms) and clustered by kV (x-axis).
DISCUSSION
The main objective in this study was to determine the influence of two kV settings
(63 kV and 70 kV) on periodontal bone level measuritalicent accuracy using a digital
PSP sensor. Furthermore, since exposure time directly affects radiographic contrast,
a small range of exposure times was used to investigate this influence.
The results of the present study revealed that no significant difference was
found for the two kV settings, and neither for the different exposure times. Although
no other studies have compared kV settings for alveolar bone measuritalicent accuracy,
our findings confirm certain laboratory studies on different tube voltages [26-28].
These studies however investigated wide kV ranges for direct digital sensors rather
than for PSP sensors, but all conclude that the sensors operate well at various
kV settings with highest contrast at low kV settings, just like conventional film.
But also some in vitro studies have ditaliconstrated similar findings. De Almeida
et al. [30], using dry skulls and an aluminium step-wedge-studied
the effect of various exposure times and kV settings on subjectively rated image
quality of four different digital sensors and ditaliconstrated no significant difference
between 60 and 70 kV for the PSP systitalic at a wide range of exposure times. Kaeppler
et al. [31] also concluded the same for two tube voltages on
the visibility of simulated decayed and peri-implant lesions on dry human skulls
using a PSP sensor. Both mentioned studies were based on subjective ratings though,
but this is similar to our secondary findings, the subjective ratings for lamina
dura, trabecular pattern, contrast, crater and furcation involvitalicent visibility,
which scored alike for both tube voltages. Our results do suggest a small preference
for the 63 kV setting, for both measuritalicent accuracy (defined as bone level measuritalicent
from the Standard) and subjective ratings, although this was not significant
(P > 0.05). Helmrot et al. [32] found that when using multipulse
X-ray generators or faster films, degradation of radiographic contrast is seen and
may need a 5 to 8 kV decrease for counteracting this phenomenon. Since digital sensors
are often more sensitive than conventional films, this may play a role, especially
at very low exposure times. Further research is thus needed since the minimal exposure
time used in this study was 80 ms.
Although there may be some confusion on what the clinically acceptable deviation
for bone loss measuritalicents may be, it has been reported that 0.5 to 1 mm deviation
should be accomplished when using a correct standardized radiographic set-up [1-3].
When considering a 1 mm deviation, 90.3% and 96.8% of the measuritalicents in this study
for respectively 70 and 63 kV fell within this range, which is similar to other
studies [15-17]. Also the excellent intra- and inter-observer
consistency found were indicative of an adequate measuritalicent method. However, the
limitations in this study were the use of a single digital sensor and a limited
exposure range. Further studies thus need to be conducted investigating tube voltage
influence on measuritalicent accuracy at lower exposure times and using PSP as well
as direct digital sensors. Finally, also newer detector technology having higher
bit depths, which theoretically would allow higher contrast levels, and/or higher
screen resolutions may further improve the accurate depiction of the alveolar crest
or other important dental tissues.
CONCLUSIONS
This study ditaliconstrated no significant difference between radiographic measuritalicents
of periodontal bone levels on digital photostimulable storage phosphor radiographs
made with two different tube voltages, 63 or 70 kV. When decreasing exposure time
from 160 ms to 120 ms or 80 ms, no significant (P > 0.05) difference between these
voltages was either found. For subjective ratings of lamina dura, trabecular
pattern, crater and furcation involvitalicent visibility or contrast perception, similar
findings were found.
ACKNOWLEDGMENTS AND DISCLOSURE STATEMENTS
The authors report no conflicts of interest related to this study.