- Email: [email protected]
Radiopacity of dental materials using a digital X-ray system
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
available at www.sciencedirect.com
journal homepage: www.intl.elsevierhealth.com/journals/dema
Radiopacity of dental materials using a digital X-ray system Steven Gu ∗ , Brian J. Rasimick, Allan S. Deutsch, Barry Lee Musikant Essential Dental Laboratories, 89 Leuning Street, South Hackensack, NJ 07606, USA
a r t i c l e
i n f o
a b s t r a c t
Article history:
Objectives. Radiopacity is a desirable property for most intra-oral materials. There are estab-
Received 17 June 2005
lished ISO and ANSI/ADA protocols for determining radiopacity using film-based radiogra-
Received in revised form 29 August
phy. However, these methods are not always followed by researchers. This study aims to
2005
adapt those procedures by using digital radiography, a simplified stepwedge, and examine
Accepted 21 September 2005
the effects of target distance and exposure time choice. Methods. One millimetre thick samples of three dental materials were prepared by placing the materials into a 1.00 mm thick washer sandwiched between two glass slides. The sam-
Keywords:
ples were digitally radiographed alongside a stepwedge of aluminum alloy 1100 with an
Radiopacity
X-ray unit at 70 kVp using five different target distance/exposure time combinations. For
Radio-opacity digital radiography
each combination, the grey scale values of various thicknesses of the stepwedge were con-
Step wedge of aluminum
verted into absorbencies and plotted against their thickness. These plots were then linearly
Target distance
regressed in order to correlate absorbance with a thickness of aluminum for each target dis-
Beer’s Law
tance/exposure time combination. The absorbencies of each sample were then converted into radiopacities using these correlations. Results. The correlations between the absorbance of the stepwedge and its thickness were highly linear. This linearity allows the correlation to be accurately deduced from fewer data points than required by the ISO and ANSI/ADA protocols. Varying exposure time did not significantly affect the mean radiopacity measured at a target distance of 30 cm. Varying the target distance did not significantly affect the measured radiopacity as long as the samples were properly exposed. Significance. A simplified, consistent digital method for determining radiopacity is presented. © 2005 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Dental diagnosis relies heavily on radiology. In order to identify and distinguish an intra-oral material from surrounding anatomical structures, the material must be radiopaque. Both the ISO and ANSI/ADA have published standardized procedures for quantifying the radiopacity of several types of materials using aluminum alloy 1100 as a reference [1–5]. However,
∗
several researches have developed modified versions of the standard procedures. One of the most popular modifications is the use of a simplified aluminum stepwedge as the reference standard. Most of the ISO and ADA protocols require a stepwedge of aluminum to be machined from a single aluminum block to an accuracy of 10 m [1–3]. Such machining is expensive and not readily available. As a result, several researchers [6–10] have
Corresponding author. Tel.: +1 201 487 9090; fax: +1 201 487 5120. E-mail address: [email protected] (S. Gu).
0109-5641/$ – see front matter © 2005 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.dental.2005.11.004
766
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
tried to reduce machining costs by using a stepwedge with steps higher than the 0.5 mm recommended by some of the ISO protocols [2,3]. Stepwedges with higher (and thus fewer) steps have the added benefit of speeding the measurement process. However, it is unknown if the accuracy of the results are comparable. Low-cost stepwedges can also be fabricated by riveting several strips of aluminum alloy 1100 together rather than using a solid block [6,7]. Tagger and Katz recently suggested that the existing protocols for determining radiopacity might be improved by incorporating digital radiography [6]. In addition to reducing the operator’s exposure to radiation and eliminating the need for film development chemicals, digital radiography also provides consistent radiograph ‘development’. Traditional film development, unless performed carefully, can produce significant variations in the final radiograph [3,11]. Thus, a digital method should provide more consistent results. The main purpose of this study was to refine the existing techniques for measuring radiopacity in order to make them quicker, simpler, and more consistent. Our innovations include the use of digital radiography and a simple yet accurate calibration that employs a simplified step-wedge of aluminum alloy 1100. The refined method was then used to assess the effect of five target distance–exposure time combinations upon the measured radiopacity of three representative dental materials.
2.
Materials and methods
The materials evaluated in this study were the universal cement Embrace WetBond (Pulpdent, Watertown, MA), the universal cement RelyX UniCem Aplicap shade A2 Universal (3M EPSE, Seefeld, Germany), and the root canal sealer ` RoekoSeal Auto (Coltene/Whaledent, Langenau, Germany). These materials represent a range of radiopacities of typical dental products. Ten samples of each material were analyzed. The materials were prepared according to their manufacturer’s instructions and placed into wells created by clipping 1.00 mm thick plates of aluminum alloy 1100 (>99.0 Al) (Alcoa, Pittsburg, PA) containing a 4 mm diameter hole on top of glass microscope slides. After the wells were filled, glass slides were clipped on top of the aluminum plates, covering the samples in order to create uniform 1 mm thick disks of the materials. Each sample was then digitally imaged alongside an aluminum stepwedge that was used as a reference. The stepwedge was fabricated by riveting together 15 1.00-mm thick plates of aluminum alloy 1100. The plates were 10.0 mm wide and their lengths ranged from 30.0 mm at the base of the wedge to 15.0 mm at the top. The images were taken using an RVG sensor (Trophy Radiology, Inc., Marietta, GA) and a dental X-ray machine (AcuRay 071A, Belmont, Somerset, NJ) operating at 70 kVp and 10 mA with a total filtration equivalent to 2.25 mm of aluminum. Each sample was radiographed using five different combinations of exposure time and filter-totarget distance: 40 cm 8/60 s, 30 cm 8/60 s, 30 cm 6/60 s, 30 cm 3/60 s, and 15 cm 3/60 s. The 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm 3/60 s combinations were considered properly exposed.
The 30 cm 8/60 s combination was considered to be overexposed because it could not visualize all of the stepwedge; 1 mm of aluminum alloy 1100 had a grey-scale value of 0. The 30 cm 3/60 s combination was considered to be underexposed because it produced images with a background fog. The digital images were analyzed with the Trophy software system. The ‘bone density’ tool was applied to the region of the radiographs containing the sample. Care was taken to analyze only those regions, which were free of air bubbles and other anomalies. The bone density tool produced a graph of the grey-scale value of each pixel (0 (black) to 255 (white)) in the analyzed segment. The grey-scale value corresponds to the attenuation of the material. From the graph, the average attenuation of the region was estimated to ±2 (the resolution of the graph) and converted into an absorbance by using the following formula:
A = −log 10 (T) = −log 10 1 −
G 255
where A is the absorbance, T is the transmittance, and G is the grey-scale value of the item. For each of the five exposure time/target distance combinations, 10 out of the 30 radiographs were chosen at random. In each radiograph, the bone density tool was applied to the regions containing the first nine steps of the aluminum stepwedge. The grey-scale data for each exposure time/target distance combination were converted to absorbencies and plotted against the number of aluminum steps. The plots were then linearly regressed, and the regressions were used to correlate absorbance with aluminum alloy 1100 thickness. These correlations were used to convert the previously recorded absorbencies of the material/slide samples into thicknesses of aluminum alloy 1100. The 10 radiographs were also analyzed to determine the absorbance of 1.00 mm of aluminum alloy 1100 (the plate) and the two glass slides. The mean radiopacities of the slides were subtracted from the sample data taken at the corresponding target distance in order to find the radiopacity of the materials. The experiment was not a complete factorial design. Therefore, three-way ANOVA could not test for interactions involving exposure time and distance. A separate two-way ANOVA was preformed on material and target distance/exposure groupings of 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm 3/60 s (properly exposed combinations). Another two-way ANOVA was preformed on material and exposure time groupings of 30 cm 8/60 s, 30 cm 6/60 s, and 30 cm 3/60 s (over-exposed, properly exposed, and under-exposed). Results were considered significant if p < 0.05.
3.
Results
The RGV sensor when covered with a lead sheet gave a greyscale value of 255. This confirms that the constant used in the equation given in the introduction of this paper, 255, is appropriate. The absorbance of the aluminum alloy 1100 step-wedge at a target distance of 30 cm is plotted in Fig. 1 for three different exposure times. The plot also depicts the least-squares linear
767
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
Table 1 – Absorbance/thickness correlation regression parameters for the equation a × x + b and associated errors Conditions
Regression parameter ‘a’ fit (standard error)
Regression parameter ‘b’ fit (standard error)
R2
Average magnitude of the mean residuals (mm aluminum 1100)
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
0.0996 (0.0006) 0.1014 (0.0005) 0.1023 (0.0006) 0.0944 (0.0011) 0.1040 (0.0005)
−0.0122 (0.0035) −0.1495 (0.0029) −0.0528 (0.0035) +0.2435 (0.0059) −0.0680 (0.0029)
0.9965 0.9983 0.9967 0.9891 0.9978
0.051 0.024 0.047 0.076 0.047
Fig. 1 – Absorbance of the aluminum step-wedge exposed for 3/60 s (—), 6/60 s (– –), and 8/60 s (- - -) at target distance of 30 cm.
regressions of the data sets. The equations of the best-fit lines and their associated errors are given in Table 1. The average magnitude of the mean regression residuals was 0.049 mm of aluminum and the maximum regression residual was −0.134 mm of aluminum alloy 1100. The residuals were random with respect to radiopacity, indicating that no major non-linearities were present. Table 2 lists the equations of the best-fit lines calculated from only from the data for steps 2, 5, and 8 of the stepwedge. It also lists the errors of the equations. Table 3 and Fig. 2 list the radiopacity of the materials under the five exposure time/target distance combinations. Two-way ANOVA found no significant difference between the mean radiopacities measured at a target distance of 30 cm using the three different exposure times (p = 0.11). However, the standard deviation of the measurements made from the under-exposed images appeared to be larger. Two-way ANOVA also indicated no significant difference between the mean radiopacities measured at the exposure time/target distance combinations that are considered clinically relevant, 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm, 3/60 s (p = 0.24). The standard deviations of the means appeared to be similar.
Table 2 – Absorbance/thickness correlation regression parameters for the equation a × x + b using only the data from steps 2, 5, and 8 of the stepwedge Conditions
Regression parameter ‘a’ fit (standard error)
Regression parameter ‘b’ fit (standard error)
R2
Average magnitude of the mean residuals (mm aluminum 1100)a
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
0.0993 (0.0011) 0.1011 (0.0007) 0.1017 (0.0012) 0.0937 (0.0019) 0.1036 (0.0008)
−0.0081 (0.0059) −0.1502 (0.0041) −0.0503 (0.0067) +0.2470 (0.0105) −0.0644 (0.0046)
0.9968 0.9985 0.9960 0.9887 0.9982
0.053 0.027 0.047 0.076 0.047
a
Residuals were calculated using all of the stepwedge data.
Table 3 – Radiopacity of the materials measured under the five exposure time/target distance combinations Conditions
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
Mean radiopacity in mm of aluminum alloy 1100 Embrace
RelyX UniCem
1.28 (0.09) 1.23 (0.08) 1.24 (0.08) 1.35 (0.16) 1.28 (0.11)
2.60 (0.13) 2.64 (0.09) 2.66 (0.11) 2.52 (0.17) 2.72 (0.09)
RoekoSeal Auto 5.27 (0.18) 5.35 (0.16) 5.13 (0.14) 5.02 (0.49) 5.19 (0.13)
768
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
Fig. 2 – Radiopacity of Embrace WetBond (), RelyX UniCem ( ), and RoekoSeal Auto ( ) measured under the five exposure time/target distance combinations.
4.
Discussion
The three materials used in this study were chosen in order to investigate a range of radiopacities commonly encountered in dental materials. Of all the ISO and ANSI/ADA requirements for dental materials, the lowest radiopacity requirement is 1 mm of aluminum alloy 1100 per mm of the material [3]. Embrace WetBond is slightly more radiopaque than this minimum. RelyX UniCem is twice as radiopaque as Embrace, and RoekoSeal Auto is twice as radiopaque as RelyX UniCem. Thus, the protocol presented in this paper has been verified for materials with equivalent radiopacities between 1 and 5.5 mm aluminum alloy 1100. However, some endodontic sealants are extremely radiopaque, even as much as 11 mm aluminum alloy 1100 per mm of material [6]. Future studies will investigate if the method presented in this paper can be applied to those materials. All materials investigated in this study exceeded the minimum required radiopacities stipulated by the ISO and ANSI/ADA. The rationale behind these requirements is outside the scope of this paper. The Beer–Lambert Law, a concept often used in spectroscopy, suggests that the correlation between absorbance and aluminum alloy 1100 thickness should be linear for monochromatic X-rays. Although the bremsstrahlung radiation initially produced by the X-ray unit is polychromatic, the total filtration (inherent and added) of the X-ray unit used in this study, 2.25 mm Al, decreases the effective bandwidth of the radiation emitted from the tube. A theoretical correlation between absorbance and pure aluminum thickness (0–9 mm) for the polychromatic radiation can be calculated using X-ray attenuation data provided by the NIST [12] and making several assumptions. First, assume that the intensity of the X-rays emitted by the anode as a function of energy is a line with neagative slope. This has been reported in literature [13]. Next, assume that the photon production of X-ray sensor scintillator is directly proportional to the energy of the incident X-ray and that the CCD or CMOS captures all photons produced by
the scintillation crystal equally well. This is a good assumption because the light output of most scintillation crystals is reported as constant number of photons per keV of the incident photon and the output wavelength of the scintillator is chosen to coincide with the detection range of the photosensor. Because the intensity of the incident radiation is equal to the number of photons multiplied by their energy, the photon output of the scintilator crystal is directly proportional to intensity. The calculated theoretical correlation is approximately linear with an R2 -value of 0.996 and an average residual magnitude of 0.14 mm of aluminum. These residuals were not random. In fact, a second degree polynomial with minimal curvature fit the theoretical data better (R2 = 0.9999). Thus, the highly linear correlations observed in this experiment (R2 between 0.989 and 0.998) are expected. The nearly perfect linearity allows one to obtain an accurate linear regression from fewer data points. For example, Table 3 shows that using only the data from the second, fifth, and eighth steps of the aluminum alloy 1100 stepwedge produced regressions that were similar to those created from the full set of stepwedge data. All of the slopes and three of the five intercepts fell within the standard error of the parameters calculated from the full set of stepwedge data. The two that did not fell within 1.3 standard errors. Using only the data from the second, fifth, and eighth step only increase the average magnitude of the mean residuals by 0.003 mm of aluminum alloy 1100 at the most. Therefore, a reference block with fewer steps could be used without sacrificing much accuracy. This stepwedge would be cheaper to construct and allow for a much quicker calibration. Any errors created by using the simplified regression are especially negligible in light of the fact that the exact elemental composition of the stepwedge will influence the measured radiopacity [14,15]. The least stringent ISO dental protocols [2,3] require aluminum of at least 98.0% purity with no more than 1% iron and 0.1% copper. A survey of 12 stepwedges used in industry and academia found that only 60% of them met these modest requirements [15]. It has been shown that using a stepwedge of an aluminum alloy with 4% copper will lead to radiopacity measurements a full 50% lower than ones taken with 99.5% aluminum [14]. A 0.1% copper should therefore create a 1.25% systematic error. Like copper, iron also attenuates X-rays more strongly than aluminum and its presence in the stepwedge will cause lowered radiopacity readings. Thus, in order to compare radiopacity measurements done by different researchers, it is imperative that all radiopacity measurements are taken with a standardized grade of aluminum. This paper utilized aluminum alloy 1100 because it has been suggested in three of the dental protocols [1,4,5]. Similar theoretical calculations can also be preformed for radiopacifiers commonly used in dental products such as barium, bismuth, tungsten, and zirconium. The correlations for these materials are also approximately linear (R2 of 0.998, 0.993, 0.993, and 0.992, average residual magnitude of 0.11, 0.24, 0.24, and 0.26 mm aluminum respectively) [12] for radiopacities less than 9 mm of aluminum alloy 1100 per mm of material. This suggests that for dental materials containing those radiopacifiers, the measured radiopacity per millimeter of material should not be appreciably affected by the thickness of the sample being measured.
dental materials
Digital radiology does not involve film development, a process that introduces variation in the final radiograph [3,8,11]. Thus, the absorbance of the aluminum alloy 1100 stepwedge changes very little between digital radiographs taken at the same exposure time and target distance. As a result, if a digital technique is used, it is unnecessary to measure the absorbance of the stepwedge in every radiograph so long as the target distance and exposure time remain unchanged. Although the initial choice of exposure time and target distance had no effect on the measured radiopacity, changing their values during the experiment will. Thus, the chosen target distance and exposure time must be reproducible. A convenient way to ensure this is to use the X-ray head cone as a guide and place the sample to be analyzed right at the end of or just inside of the cone. From a theoretical standpoint, long target distances are suggested help to ensure that the X-ray sensor or film is uniformly irradiated. When the X-ray filament is far away from the X-ray sensor, it can be assumed that all points on the X-ray sensor lie an equal distance away from the filament. As the target distance approaches 0, this assumption clearly breaks down. Short target distances should also cause the ‘heel effect,’ the phenomena by which more X-rays are produced in the direction of the cathode than the anode, to have a more pronounced effect on the radiographs. Finally, as the target distance decreases, the X-rays in the beam striking the sensor become less parallel to each other. This can cause the resulting radiograph to be blurred. The results of this study indicate that a target distance of 15 cm produces radiopacity values that are statistically no different from those measured using a target distance of 30 or 40 cm. Furthermore, the precision of the measurements, as represented by the standard deviations of the measurements, also remained unchanged. Thus, the different target distance stipulations made in various ISO protocols for dental products (40 cm for ISO 4049 [3], 30 cm in the other protocols [1,2,4,5]) seem irrelevant. The high voltage used to accelerate the electrons, which ultimately produce the X-rays may vary over the first few moments of X-ray production. Obviously the high-voltage cannot be applied truly instantaneously. Designers of X-ray generators have undoubted taken great care to minimize the time required to establish a stable full voltage. However, if the time is non-negligable compared to the short exposure times used in digital radiography, then the X-ray beam during the initial moments of the exposure will be composed of lower energy photons that attenuate differently than the full energy beam. The results of this study indicated that this effect was not noticeable when using a target distance of 30 cm. However, the standard deviation of the measurements for under-exposed images was, on average, 2.5 times greater than that of the value determined using properly exposed or over-exposed radiographs. An underexposed image has a background fog and hence can only visualize objects using a fraction of the grey-scale spectrum. This results in radiopacity measurements of decreased precision. For example, objects imaged using a 3/60 s exposure and 30 cm target distance were represented on a scale of approximately 105–255 rather than the full 0–255. The optimal exposure time for a given target distance should maximize the grey-scale range of the objects of inter-
22
( 2 0 0 6 ) 765–770
769
est in the radiograph. For the purposes of this experiment, the time had to be long enough not to produce much of a background fog, but short enough to visualize 1 mm of aluminum alloy 1100. An underexposed image has a background fog. Overexposed images ‘black out’ objects of low radiopacity. The target distance/exposure time combinations that were considered optimal for this experiment were 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm, 3/60 s. However, these optimums depend on the exact equipment used and its condition. The Trophy sensor used in this experiment is similar to most other commercial digital intraoral sensors in that it uses a scintillation screen to convert incident X-rays into less energetic radiation that is then detected by another sensor such as a CCD or CMOS. Thus, the results of this experiment are expected to be applicable to most digital sensor systems. However, as of now, there is no experimental evidence to support this generalization. To summarize, the observed linearity of the absorbance of the aluminum stepwedge as a function of thickness allows for the use of a simple three level stepwedge that is both simple to construct and convenient to use. Because digital radiology produces consistent exposures, the absorbance of the stepwedge does not have to be measured in every radiograph so long as the exposure time and target distance are kept constant. The theoretical linearity of the absorbance of radiopacifing agents as a function of thickness suggests that the measured radiopacity per millimeter of material should not be affected by the thickness of the sample being measured. Finally, the measured mean radiopacity depended on the initial choice of neither exposure time nor target distance. However, choosing a proper exposure time can reduce the standard deviation of the measurements.
references
[1] ANSI/ADA, American National Standards Institue/American Dental Association. Specification No. 57, Endodontic Sealing Material, 2000. [2] International Standards Organization. ISO 6876. Dental Root Canal Sealing Materials, 2nd ed., 2001. [3] International Standards Organization. ISO 4049. Dentistry—Polymer-Based Filling, Restorative and Luting Materials, 3rd ed., 2000. [4] ANSI/ADA, American National Standards Institue/American Dental Association. Specification No. 78, Dental Obturating Cones. 2000. [5] International Standards Organization. ISO 6877. Dental Root-Canal Obturating Points, 1st ed., 1995. [6] Tagger M, Katz A. Radiopacity of endodontic sealers: development of a new method for direct measurement. J Endodon 2003;29:751–5. [7] Tagger M, Katz A. A standard for radiopacity of root-end (retrograde) filling materials is urgently needed. Int Endod J 2004;37:260–4. [8] Beyer-Olsen EM, Ørstavik D. Radiopacity of root canal sealers. Oral Surg Oral Med Oral Pathol 1981;51:320–8. [9] Sidhu SK, Shah PMM, Chong BS, Pitt Ford TR. Radiopacity of resin-modified glass-ionomer restorative cements. Quintessence Int 1996;27:639–43. [10] Shah PMM, Sidhu SK, Chong BS, Pitt Ford TR. Radiopacity of resin-modified glass ionomer liners and bases. J Prosthetic Dent 1997;77:239–42.
770
dental materials
[11] Camps J, Pommel L, Bukiet F. Evaluation of periapical lesion healing by correction of gray values. J Endodon 2004;30:762–6. [12] Chantler CT, Olsen K, Dragoset RA, Kishore AR, Kotochigova SA, Zucker DS. (2005). X-Ray Form Factor, Attenuation and Scattering Tables (version 2.1). [Online] Available: http://physics.nist.gov/ffast. National Institute of Standards and Technology, Gaithersburg, MD; Originally published as Chantler CT. J Phys Chem Ref Data 2000;29(4):597–1048; Chantler CT. J Phys Chem Ref Data 1995;24:71–643.
22
( 2 0 0 6 ) 765–770
[13] “Bremsstralhung.” The Encyclopaedia of Medical Imaging. Volume I. [Online] Available: http://amershamhealth.com/medcyclopaedia/medical/. Amersham Health. [14] Watts DC, McCabe JF. Aluminum radiopacity standards for dentistry: and international survey. J Dent 1999;27:73–8. [15] Watts DC. Characterization of aluminum radiopacity standards for restorative materials. J Dent 1987;15: 175–7.
available at www.sciencedirect.com
journal homepage: www.intl.elsevierhealth.com/journals/dema
Radiopacity of dental materials using a digital X-ray system Steven Gu ∗ , Brian J. Rasimick, Allan S. Deutsch, Barry Lee Musikant Essential Dental Laboratories, 89 Leuning Street, South Hackensack, NJ 07606, USA
a r t i c l e
i n f o
a b s t r a c t
Article history:
Objectives. Radiopacity is a desirable property for most intra-oral materials. There are estab-
Received 17 June 2005
lished ISO and ANSI/ADA protocols for determining radiopacity using film-based radiogra-
Received in revised form 29 August
phy. However, these methods are not always followed by researchers. This study aims to
2005
adapt those procedures by using digital radiography, a simplified stepwedge, and examine
Accepted 21 September 2005
the effects of target distance and exposure time choice. Methods. One millimetre thick samples of three dental materials were prepared by placing the materials into a 1.00 mm thick washer sandwiched between two glass slides. The sam-
Keywords:
ples were digitally radiographed alongside a stepwedge of aluminum alloy 1100 with an
Radiopacity
X-ray unit at 70 kVp using five different target distance/exposure time combinations. For
Radio-opacity digital radiography
each combination, the grey scale values of various thicknesses of the stepwedge were con-
Step wedge of aluminum
verted into absorbencies and plotted against their thickness. These plots were then linearly
Target distance
regressed in order to correlate absorbance with a thickness of aluminum for each target dis-
Beer’s Law
tance/exposure time combination. The absorbencies of each sample were then converted into radiopacities using these correlations. Results. The correlations between the absorbance of the stepwedge and its thickness were highly linear. This linearity allows the correlation to be accurately deduced from fewer data points than required by the ISO and ANSI/ADA protocols. Varying exposure time did not significantly affect the mean radiopacity measured at a target distance of 30 cm. Varying the target distance did not significantly affect the measured radiopacity as long as the samples were properly exposed. Significance. A simplified, consistent digital method for determining radiopacity is presented. © 2005 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
Dental diagnosis relies heavily on radiology. In order to identify and distinguish an intra-oral material from surrounding anatomical structures, the material must be radiopaque. Both the ISO and ANSI/ADA have published standardized procedures for quantifying the radiopacity of several types of materials using aluminum alloy 1100 as a reference [1–5]. However,
∗
several researches have developed modified versions of the standard procedures. One of the most popular modifications is the use of a simplified aluminum stepwedge as the reference standard. Most of the ISO and ADA protocols require a stepwedge of aluminum to be machined from a single aluminum block to an accuracy of 10 m [1–3]. Such machining is expensive and not readily available. As a result, several researchers [6–10] have
Corresponding author. Tel.: +1 201 487 9090; fax: +1 201 487 5120. E-mail address: [email protected] (S. Gu).
0109-5641/$ – see front matter © 2005 Academy of Dental Materials. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.dental.2005.11.004
766
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
tried to reduce machining costs by using a stepwedge with steps higher than the 0.5 mm recommended by some of the ISO protocols [2,3]. Stepwedges with higher (and thus fewer) steps have the added benefit of speeding the measurement process. However, it is unknown if the accuracy of the results are comparable. Low-cost stepwedges can also be fabricated by riveting several strips of aluminum alloy 1100 together rather than using a solid block [6,7]. Tagger and Katz recently suggested that the existing protocols for determining radiopacity might be improved by incorporating digital radiography [6]. In addition to reducing the operator’s exposure to radiation and eliminating the need for film development chemicals, digital radiography also provides consistent radiograph ‘development’. Traditional film development, unless performed carefully, can produce significant variations in the final radiograph [3,11]. Thus, a digital method should provide more consistent results. The main purpose of this study was to refine the existing techniques for measuring radiopacity in order to make them quicker, simpler, and more consistent. Our innovations include the use of digital radiography and a simple yet accurate calibration that employs a simplified step-wedge of aluminum alloy 1100. The refined method was then used to assess the effect of five target distance–exposure time combinations upon the measured radiopacity of three representative dental materials.
2.
Materials and methods
The materials evaluated in this study were the universal cement Embrace WetBond (Pulpdent, Watertown, MA), the universal cement RelyX UniCem Aplicap shade A2 Universal (3M EPSE, Seefeld, Germany), and the root canal sealer ` RoekoSeal Auto (Coltene/Whaledent, Langenau, Germany). These materials represent a range of radiopacities of typical dental products. Ten samples of each material were analyzed. The materials were prepared according to their manufacturer’s instructions and placed into wells created by clipping 1.00 mm thick plates of aluminum alloy 1100 (>99.0 Al) (Alcoa, Pittsburg, PA) containing a 4 mm diameter hole on top of glass microscope slides. After the wells were filled, glass slides were clipped on top of the aluminum plates, covering the samples in order to create uniform 1 mm thick disks of the materials. Each sample was then digitally imaged alongside an aluminum stepwedge that was used as a reference. The stepwedge was fabricated by riveting together 15 1.00-mm thick plates of aluminum alloy 1100. The plates were 10.0 mm wide and their lengths ranged from 30.0 mm at the base of the wedge to 15.0 mm at the top. The images were taken using an RVG sensor (Trophy Radiology, Inc., Marietta, GA) and a dental X-ray machine (AcuRay 071A, Belmont, Somerset, NJ) operating at 70 kVp and 10 mA with a total filtration equivalent to 2.25 mm of aluminum. Each sample was radiographed using five different combinations of exposure time and filter-totarget distance: 40 cm 8/60 s, 30 cm 8/60 s, 30 cm 6/60 s, 30 cm 3/60 s, and 15 cm 3/60 s. The 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm 3/60 s combinations were considered properly exposed.
The 30 cm 8/60 s combination was considered to be overexposed because it could not visualize all of the stepwedge; 1 mm of aluminum alloy 1100 had a grey-scale value of 0. The 30 cm 3/60 s combination was considered to be underexposed because it produced images with a background fog. The digital images were analyzed with the Trophy software system. The ‘bone density’ tool was applied to the region of the radiographs containing the sample. Care was taken to analyze only those regions, which were free of air bubbles and other anomalies. The bone density tool produced a graph of the grey-scale value of each pixel (0 (black) to 255 (white)) in the analyzed segment. The grey-scale value corresponds to the attenuation of the material. From the graph, the average attenuation of the region was estimated to ±2 (the resolution of the graph) and converted into an absorbance by using the following formula:
A = −log 10 (T) = −log 10 1 −
G 255
where A is the absorbance, T is the transmittance, and G is the grey-scale value of the item. For each of the five exposure time/target distance combinations, 10 out of the 30 radiographs were chosen at random. In each radiograph, the bone density tool was applied to the regions containing the first nine steps of the aluminum stepwedge. The grey-scale data for each exposure time/target distance combination were converted to absorbencies and plotted against the number of aluminum steps. The plots were then linearly regressed, and the regressions were used to correlate absorbance with aluminum alloy 1100 thickness. These correlations were used to convert the previously recorded absorbencies of the material/slide samples into thicknesses of aluminum alloy 1100. The 10 radiographs were also analyzed to determine the absorbance of 1.00 mm of aluminum alloy 1100 (the plate) and the two glass slides. The mean radiopacities of the slides were subtracted from the sample data taken at the corresponding target distance in order to find the radiopacity of the materials. The experiment was not a complete factorial design. Therefore, three-way ANOVA could not test for interactions involving exposure time and distance. A separate two-way ANOVA was preformed on material and target distance/exposure groupings of 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm 3/60 s (properly exposed combinations). Another two-way ANOVA was preformed on material and exposure time groupings of 30 cm 8/60 s, 30 cm 6/60 s, and 30 cm 3/60 s (over-exposed, properly exposed, and under-exposed). Results were considered significant if p < 0.05.
3.
Results
The RGV sensor when covered with a lead sheet gave a greyscale value of 255. This confirms that the constant used in the equation given in the introduction of this paper, 255, is appropriate. The absorbance of the aluminum alloy 1100 step-wedge at a target distance of 30 cm is plotted in Fig. 1 for three different exposure times. The plot also depicts the least-squares linear
767
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
Table 1 – Absorbance/thickness correlation regression parameters for the equation a × x + b and associated errors Conditions
Regression parameter ‘a’ fit (standard error)
Regression parameter ‘b’ fit (standard error)
R2
Average magnitude of the mean residuals (mm aluminum 1100)
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
0.0996 (0.0006) 0.1014 (0.0005) 0.1023 (0.0006) 0.0944 (0.0011) 0.1040 (0.0005)
−0.0122 (0.0035) −0.1495 (0.0029) −0.0528 (0.0035) +0.2435 (0.0059) −0.0680 (0.0029)
0.9965 0.9983 0.9967 0.9891 0.9978
0.051 0.024 0.047 0.076 0.047
Fig. 1 – Absorbance of the aluminum step-wedge exposed for 3/60 s (—), 6/60 s (– –), and 8/60 s (- - -) at target distance of 30 cm.
regressions of the data sets. The equations of the best-fit lines and their associated errors are given in Table 1. The average magnitude of the mean regression residuals was 0.049 mm of aluminum and the maximum regression residual was −0.134 mm of aluminum alloy 1100. The residuals were random with respect to radiopacity, indicating that no major non-linearities were present. Table 2 lists the equations of the best-fit lines calculated from only from the data for steps 2, 5, and 8 of the stepwedge. It also lists the errors of the equations. Table 3 and Fig. 2 list the radiopacity of the materials under the five exposure time/target distance combinations. Two-way ANOVA found no significant difference between the mean radiopacities measured at a target distance of 30 cm using the three different exposure times (p = 0.11). However, the standard deviation of the measurements made from the under-exposed images appeared to be larger. Two-way ANOVA also indicated no significant difference between the mean radiopacities measured at the exposure time/target distance combinations that are considered clinically relevant, 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm, 3/60 s (p = 0.24). The standard deviations of the means appeared to be similar.
Table 2 – Absorbance/thickness correlation regression parameters for the equation a × x + b using only the data from steps 2, 5, and 8 of the stepwedge Conditions
Regression parameter ‘a’ fit (standard error)
Regression parameter ‘b’ fit (standard error)
R2
Average magnitude of the mean residuals (mm aluminum 1100)a
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
0.0993 (0.0011) 0.1011 (0.0007) 0.1017 (0.0012) 0.0937 (0.0019) 0.1036 (0.0008)
−0.0081 (0.0059) −0.1502 (0.0041) −0.0503 (0.0067) +0.2470 (0.0105) −0.0644 (0.0046)
0.9968 0.9985 0.9960 0.9887 0.9982
0.053 0.027 0.047 0.076 0.047
a
Residuals were calculated using all of the stepwedge data.
Table 3 – Radiopacity of the materials measured under the five exposure time/target distance combinations Conditions
40 cm, 8/60 s 30 cm, 8/60 s 30 cm, 6/60 s 30 cm, 3/60 s 15 cm, 3/60 s
Mean radiopacity in mm of aluminum alloy 1100 Embrace
RelyX UniCem
1.28 (0.09) 1.23 (0.08) 1.24 (0.08) 1.35 (0.16) 1.28 (0.11)
2.60 (0.13) 2.64 (0.09) 2.66 (0.11) 2.52 (0.17) 2.72 (0.09)
RoekoSeal Auto 5.27 (0.18) 5.35 (0.16) 5.13 (0.14) 5.02 (0.49) 5.19 (0.13)
768
d e n t a l m a t e r i a l s 2 2 ( 2 0 0 6 ) 765–770
Fig. 2 – Radiopacity of Embrace WetBond (), RelyX UniCem ( ), and RoekoSeal Auto ( ) measured under the five exposure time/target distance combinations.
4.
Discussion
The three materials used in this study were chosen in order to investigate a range of radiopacities commonly encountered in dental materials. Of all the ISO and ANSI/ADA requirements for dental materials, the lowest radiopacity requirement is 1 mm of aluminum alloy 1100 per mm of the material [3]. Embrace WetBond is slightly more radiopaque than this minimum. RelyX UniCem is twice as radiopaque as Embrace, and RoekoSeal Auto is twice as radiopaque as RelyX UniCem. Thus, the protocol presented in this paper has been verified for materials with equivalent radiopacities between 1 and 5.5 mm aluminum alloy 1100. However, some endodontic sealants are extremely radiopaque, even as much as 11 mm aluminum alloy 1100 per mm of material [6]. Future studies will investigate if the method presented in this paper can be applied to those materials. All materials investigated in this study exceeded the minimum required radiopacities stipulated by the ISO and ANSI/ADA. The rationale behind these requirements is outside the scope of this paper. The Beer–Lambert Law, a concept often used in spectroscopy, suggests that the correlation between absorbance and aluminum alloy 1100 thickness should be linear for monochromatic X-rays. Although the bremsstrahlung radiation initially produced by the X-ray unit is polychromatic, the total filtration (inherent and added) of the X-ray unit used in this study, 2.25 mm Al, decreases the effective bandwidth of the radiation emitted from the tube. A theoretical correlation between absorbance and pure aluminum thickness (0–9 mm) for the polychromatic radiation can be calculated using X-ray attenuation data provided by the NIST [12] and making several assumptions. First, assume that the intensity of the X-rays emitted by the anode as a function of energy is a line with neagative slope. This has been reported in literature [13]. Next, assume that the photon production of X-ray sensor scintillator is directly proportional to the energy of the incident X-ray and that the CCD or CMOS captures all photons produced by
the scintillation crystal equally well. This is a good assumption because the light output of most scintillation crystals is reported as constant number of photons per keV of the incident photon and the output wavelength of the scintillator is chosen to coincide with the detection range of the photosensor. Because the intensity of the incident radiation is equal to the number of photons multiplied by their energy, the photon output of the scintilator crystal is directly proportional to intensity. The calculated theoretical correlation is approximately linear with an R2 -value of 0.996 and an average residual magnitude of 0.14 mm of aluminum. These residuals were not random. In fact, a second degree polynomial with minimal curvature fit the theoretical data better (R2 = 0.9999). Thus, the highly linear correlations observed in this experiment (R2 between 0.989 and 0.998) are expected. The nearly perfect linearity allows one to obtain an accurate linear regression from fewer data points. For example, Table 3 shows that using only the data from the second, fifth, and eighth steps of the aluminum alloy 1100 stepwedge produced regressions that were similar to those created from the full set of stepwedge data. All of the slopes and three of the five intercepts fell within the standard error of the parameters calculated from the full set of stepwedge data. The two that did not fell within 1.3 standard errors. Using only the data from the second, fifth, and eighth step only increase the average magnitude of the mean residuals by 0.003 mm of aluminum alloy 1100 at the most. Therefore, a reference block with fewer steps could be used without sacrificing much accuracy. This stepwedge would be cheaper to construct and allow for a much quicker calibration. Any errors created by using the simplified regression are especially negligible in light of the fact that the exact elemental composition of the stepwedge will influence the measured radiopacity [14,15]. The least stringent ISO dental protocols [2,3] require aluminum of at least 98.0% purity with no more than 1% iron and 0.1% copper. A survey of 12 stepwedges used in industry and academia found that only 60% of them met these modest requirements [15]. It has been shown that using a stepwedge of an aluminum alloy with 4% copper will lead to radiopacity measurements a full 50% lower than ones taken with 99.5% aluminum [14]. A 0.1% copper should therefore create a 1.25% systematic error. Like copper, iron also attenuates X-rays more strongly than aluminum and its presence in the stepwedge will cause lowered radiopacity readings. Thus, in order to compare radiopacity measurements done by different researchers, it is imperative that all radiopacity measurements are taken with a standardized grade of aluminum. This paper utilized aluminum alloy 1100 because it has been suggested in three of the dental protocols [1,4,5]. Similar theoretical calculations can also be preformed for radiopacifiers commonly used in dental products such as barium, bismuth, tungsten, and zirconium. The correlations for these materials are also approximately linear (R2 of 0.998, 0.993, 0.993, and 0.992, average residual magnitude of 0.11, 0.24, 0.24, and 0.26 mm aluminum respectively) [12] for radiopacities less than 9 mm of aluminum alloy 1100 per mm of material. This suggests that for dental materials containing those radiopacifiers, the measured radiopacity per millimeter of material should not be appreciably affected by the thickness of the sample being measured.
dental materials
Digital radiology does not involve film development, a process that introduces variation in the final radiograph [3,8,11]. Thus, the absorbance of the aluminum alloy 1100 stepwedge changes very little between digital radiographs taken at the same exposure time and target distance. As a result, if a digital technique is used, it is unnecessary to measure the absorbance of the stepwedge in every radiograph so long as the target distance and exposure time remain unchanged. Although the initial choice of exposure time and target distance had no effect on the measured radiopacity, changing their values during the experiment will. Thus, the chosen target distance and exposure time must be reproducible. A convenient way to ensure this is to use the X-ray head cone as a guide and place the sample to be analyzed right at the end of or just inside of the cone. From a theoretical standpoint, long target distances are suggested help to ensure that the X-ray sensor or film is uniformly irradiated. When the X-ray filament is far away from the X-ray sensor, it can be assumed that all points on the X-ray sensor lie an equal distance away from the filament. As the target distance approaches 0, this assumption clearly breaks down. Short target distances should also cause the ‘heel effect,’ the phenomena by which more X-rays are produced in the direction of the cathode than the anode, to have a more pronounced effect on the radiographs. Finally, as the target distance decreases, the X-rays in the beam striking the sensor become less parallel to each other. This can cause the resulting radiograph to be blurred. The results of this study indicate that a target distance of 15 cm produces radiopacity values that are statistically no different from those measured using a target distance of 30 or 40 cm. Furthermore, the precision of the measurements, as represented by the standard deviations of the measurements, also remained unchanged. Thus, the different target distance stipulations made in various ISO protocols for dental products (40 cm for ISO 4049 [3], 30 cm in the other protocols [1,2,4,5]) seem irrelevant. The high voltage used to accelerate the electrons, which ultimately produce the X-rays may vary over the first few moments of X-ray production. Obviously the high-voltage cannot be applied truly instantaneously. Designers of X-ray generators have undoubted taken great care to minimize the time required to establish a stable full voltage. However, if the time is non-negligable compared to the short exposure times used in digital radiography, then the X-ray beam during the initial moments of the exposure will be composed of lower energy photons that attenuate differently than the full energy beam. The results of this study indicated that this effect was not noticeable when using a target distance of 30 cm. However, the standard deviation of the measurements for under-exposed images was, on average, 2.5 times greater than that of the value determined using properly exposed or over-exposed radiographs. An underexposed image has a background fog and hence can only visualize objects using a fraction of the grey-scale spectrum. This results in radiopacity measurements of decreased precision. For example, objects imaged using a 3/60 s exposure and 30 cm target distance were represented on a scale of approximately 105–255 rather than the full 0–255. The optimal exposure time for a given target distance should maximize the grey-scale range of the objects of inter-
22
( 2 0 0 6 ) 765–770
769
est in the radiograph. For the purposes of this experiment, the time had to be long enough not to produce much of a background fog, but short enough to visualize 1 mm of aluminum alloy 1100. An underexposed image has a background fog. Overexposed images ‘black out’ objects of low radiopacity. The target distance/exposure time combinations that were considered optimal for this experiment were 40 cm 8/60 s, 30 cm 6/60 s, and 15 cm, 3/60 s. However, these optimums depend on the exact equipment used and its condition. The Trophy sensor used in this experiment is similar to most other commercial digital intraoral sensors in that it uses a scintillation screen to convert incident X-rays into less energetic radiation that is then detected by another sensor such as a CCD or CMOS. Thus, the results of this experiment are expected to be applicable to most digital sensor systems. However, as of now, there is no experimental evidence to support this generalization. To summarize, the observed linearity of the absorbance of the aluminum stepwedge as a function of thickness allows for the use of a simple three level stepwedge that is both simple to construct and convenient to use. Because digital radiology produces consistent exposures, the absorbance of the stepwedge does not have to be measured in every radiograph so long as the exposure time and target distance are kept constant. The theoretical linearity of the absorbance of radiopacifing agents as a function of thickness suggests that the measured radiopacity per millimeter of material should not be affected by the thickness of the sample being measured. Finally, the measured mean radiopacity depended on the initial choice of neither exposure time nor target distance. However, choosing a proper exposure time can reduce the standard deviation of the measurements.
references
[1] ANSI/ADA, American National Standards Institue/American Dental Association. Specification No. 57, Endodontic Sealing Material, 2000. [2] International Standards Organization. ISO 6876. Dental Root Canal Sealing Materials, 2nd ed., 2001. [3] International Standards Organization. ISO 4049. Dentistry—Polymer-Based Filling, Restorative and Luting Materials, 3rd ed., 2000. [4] ANSI/ADA, American National Standards Institue/American Dental Association. Specification No. 78, Dental Obturating Cones. 2000. [5] International Standards Organization. ISO 6877. Dental Root-Canal Obturating Points, 1st ed., 1995. [6] Tagger M, Katz A. Radiopacity of endodontic sealers: development of a new method for direct measurement. J Endodon 2003;29:751–5. [7] Tagger M, Katz A. A standard for radiopacity of root-end (retrograde) filling materials is urgently needed. Int Endod J 2004;37:260–4. [8] Beyer-Olsen EM, Ørstavik D. Radiopacity of root canal sealers. Oral Surg Oral Med Oral Pathol 1981;51:320–8. [9] Sidhu SK, Shah PMM, Chong BS, Pitt Ford TR. Radiopacity of resin-modified glass-ionomer restorative cements. Quintessence Int 1996;27:639–43. [10] Shah PMM, Sidhu SK, Chong BS, Pitt Ford TR. Radiopacity of resin-modified glass ionomer liners and bases. J Prosthetic Dent 1997;77:239–42.
770
dental materials
[11] Camps J, Pommel L, Bukiet F. Evaluation of periapical lesion healing by correction of gray values. J Endodon 2004;30:762–6. [12] Chantler CT, Olsen K, Dragoset RA, Kishore AR, Kotochigova SA, Zucker DS. (2005). X-Ray Form Factor, Attenuation and Scattering Tables (version 2.1). [Online] Available: http://physics.nist.gov/ffast. National Institute of Standards and Technology, Gaithersburg, MD; Originally published as Chantler CT. J Phys Chem Ref Data 2000;29(4):597–1048; Chantler CT. J Phys Chem Ref Data 1995;24:71–643.
22
( 2 0 0 6 ) 765–770
[13] “Bremsstralhung.” The Encyclopaedia of Medical Imaging. Volume I. [Online] Available: http://amershamhealth.com/medcyclopaedia/medical/. Amersham Health. [14] Watts DC, McCabe JF. Aluminum radiopacity standards for dentistry: and international survey. J Dent 1999;27:73–8. [15] Watts DC. Characterization of aluminum radiopacity standards for restorative materials. J Dent 1987;15: 175–7.