The effects of different metal posts, cements, and exposure parameters on cone-beam computed tomography artifacts

The use of cone-beam computed tomography (CBCT) technology in clinical dental practice has improved dental diagnostic imaging, allowing the assessment of unique 3-dimensional images beyond what is possible to achieve using conventional 2-dimensional dental images.1 However, CBCT has limitations related to its cone-beam projection geometry, such as noise, poor soft tissue contrast, and artifacts.1 An artifact is an undesirable phenomenon that can degrade image quality and compromise the diagnostic process and is defined as an image in the reconstructed data that is not part of the object under investigation.2, 3

The intensity of beam-hardening artifacts depends on the density of the materials present in the scanned area. Dental materials usually have high density, enabling their identification as radiopaque areas on radiographic images; therefore, they usually increase the intensity of artifacts.3 The intensity of artifacts produced by high-density materials such as metal posts is directly related to the atomic number of their constituents; higher-atomic-number materials lead to greater degradation of image quality.2, 4, 5, 6, 7 Fiber posts are low-density materials that can be used to restore teeth that need intracanal retention; however, metal cores are still used when a large amount of dentin is lost and there is no adequate anchorage for the restoration.8

The intensity of high-density material-induced artifacts can be affected by the CBCT scanner,7, 9 imaging acquisition protocol,6, 10 the position of the object within the field of view (FOV), and the presence of an exomass.11, 12, 13 For this reason, the acquisition parameters, detector type, and reconstruction algorithms are important parameters that must be considered while acquiring a scan and evaluating image quality in terms of image noise, contrast resolution, and artifacts.

Considering the influence of these factors on artifact intensity and their expression on CBCT images, the aim of the present study was to assess the intensity of artifacts produced by 2 metal posts, 2 cements, and different exposure parameters in scans using 2 CBCT units. This result would provide useful guidance for operators while selecting acquisition protocols to optimize image quality and minimize patient radiation dose.

This in vitro experimental study followed the Helsinki Declaration and was approved by the University’s Ethics and Research Committee (protocol: 39088714.2.0000.5187).

The sample consisted of 20 single-rooted human teeth (premolars) extracted for therapeutic reasons. All teeth were inspected by transillumination for the absence of root fractures and radiographed on phosphor plates (Digora Optime, Soredex, Tuusula, Finland) to exclude teeth with pulp nodules, internal and/or external root resorption, previous endodontic treatment, root canal multiplicity, root canal obliteration, root fractures, or any other anomaly. The inclusion criteria stated that all teeth should present a maximum root inclination of ≤5° according to Schneider’s method14 and have similar dimensions.

After cleaning and disinfection protocols, teeth crowns were removed at the cemento-enamel junction, and root canals were standardized using the Reciproc R50 system (VDW, München, Germany) with thermomechanical compression filling, and then prepared to two-thirds of the root length (Peeso Long Drill no. 1 and no. 2, Dentsply/Maillefer, São Paulo, SP, Brazil) for posterior metal post preparation and fitting. The sample was then divided into 2 groups: 10 teeth that received nickel-chromium posts (NiCr group) and 10 teeth that received silver-palladium posts (AgPd group). Subsequently, each group was subdivided equally according to the type of cement used for its restoration: zinc phosphate and Allcem resin cements.

Metal post preparation was performed by the direct technique (standard for metal casting made of Duralay acrylic resin) and the coronary part of the metal post was standardized using a heavy-base condensation silicone matrix (Zetaplus, Zhermack, Badia Polesine, Italy).

The Fit Cast-SB Plus alloy (Talmax Produtos de Prótese Dentária Ltda, Curitiba, PR, Brazil), was used to cast the NiCr post group. For the Ag-Pd post group, the Palad PDAG alloy (La Croix Dentales, Rio de Janeiro, RJ, Brazil) was used. Periapical radiographs were taken to validate post adaptation. Posteriorly, zinc phosphate cement (S. S. White, Lakewood, NJ, USA) and Allcem resin cement (FGM, Joinville, SC, Brazil) were used for post cementation.

A dentate dry human skull was coated with a 5-mm-thick piece of wax to simulate the interference of soft tissues in the image. Each tooth was coated with a 0.2 mm layer of wax and placed in an empty mandibular right canine socket. The skull was immersed in a foam box filled with water to further simulate soft tissue coverage.15 Each sample was scanned without the metal post, with the corresponding metal post (NiCr or AgPd), and after cementation with the chosen cement (zinc phosphate or Allcem).

CBCT scans were acquired using 2 scanners: a CS9000 3D (Carestream Dental Rochester, NY, USA), a charge-coupled device (CCD) flat panel, and an i-CAT Cone Beam 3D Imaging System scanner (Imaging Sciences International, Hatfield, PA, USA), an amorphous silicon flat panel with a CsI scintillator.

The CS9000 3D scans were obtained using 2 tube voltage settings (85 kV and 90 kV) and tube currents (6.3 mA and 10 mA). The voxel size and field of view (FOV) were fixed at 0.100 mm and 5 cm×3.75 cm, respectively. The i-CAT Cone Beam 3D Imaging System scans were obtained with fixed parameters of 120 kV, 5 mA, 0.125 mm voxel size, and 6 cm×8 cm FOV.

The resulting datasets were exported as Digital Imaging and Communications in Medicine (DICOM) files to be assessed using the corresponding CBCT software: CS9000 3D Imaging Software (Carestream Dental Rochester, NY, USA) and Xoran CAT v. 3.0.34 software (Xoran Technologies, Ann Arbor, MI, USA). A total of 300 scans were saved, each with a code corresponding to the tooth study conditions and protocol used.

For the subjective analysis of artifacts, 2 experienced oral and maxillofacial radiologists (10 years of experience) assessed all images independently. Prior to all examination sessions, verbal and practical instructions and calibration tests were performed. Adjustments of zoom, brightness, and contrast settings were left to each observer’s discretion.

In the subjective analysis, the inter-observer kappa coefficient was 0.400 for hypodense halos, 1.000 for hypodense lines, and 0.700 for hyperdense lines. The intra-observer kappa coefficient for observer 1 was 1.000 for hypodense halos, 0.700 for hypodense lines, and 0.700 for hyperdense lines. For observer 2, the intra-observer kappa coefficient was 0.550 for hypodense halos, 0.625 for hypodense lines, and 0.550 for hyperdense lines.

The scans were subjectively assessed using CS9000 3D Imaging Software (Carestream Dental Rochester, NY, USA) and Xoran CAT v. 3.0.34 software (Xoran Technologies, Ann Arbor, MI), displayed on a 24-inch Dell Ultra Sharp Color Monitor (Dell Inc, Austin, TX, USA) in a quiet, dimly lit room. A daily number of 15 volumes was suggested for evaluation with a minimum interval of 24 hours, but each evaluator could choose his or her individual limit as long as it did not exceed 50 volumes per day.

The amount of image artifacts was scored as absence (0), moderate presence (1), or high presence (2) for hypodense halos (dark bands), hypodense lines (dark lines), and hyperdense lines (bright lines).4

For objective artifact quantification, each DICOM file was assessed on the scanner’s native software. The axial images were imported into GIMP version 2.8.14 (the GIMP Team, USA, available at: http://www.gimp.org/) and the region corresponding to the whole root area was selected with the free selection tool viewed with a zoom of ×200. The resulting images were set to an 8-bit scale (256 gray levels), saved with a black and a white background in JPEG format, and imported into ImageJ (version 1.48, National Institutes of Health, Bethesda, MD, USA, http://rsb.info.nih.gov/ij/) (Fig. 1).

Fig. 1
A. Root area selection using the GIMP program. B. Selected area of the root with a black background. C. Selected area of the root with a white background.

• Click for larger image

To remove the area corresponding to the post in the passive or cemented-fit post images, which did not correspond to the artifact images, an overlap was made between the images of the empty tooth and the restored tooth. This step was performed in 3 axial images (cervical, middle and apical core) for each tooth using the ImageJ software to acquire a final image of each axial slice.

For image processing, each axial image of an empty tooth was duplicated, and the threshold tool was applied to convert it to a binary image (0 - root canal and background and 255 - dentin). To enable this process, the values were normalized to a scale where the post and background corresponded to 0 and the dentin to 1. Image subtraction was then performed using the multiplication tool in the program by subtracting the binary that had the empty conduit and dentin from the image with the post and dentin. As a result, the portion of the image corresponding to the post was excluded from the final image (Fig. 2).

Fig. 2
Image processing for nucleus removal. A. Image of the restored tooth. B. Image of the unrestored tooth. C. Duplicate image of the empty tooth with threshold tool. D. binary image of the empty tooth. E. subtracted image of the restored tooth

• Click for larger image

When analyzing the dentin area, the threshold tool was used to determine the hypodense and hyperdense artifact areas. In an 8-bit image, with 256 gray levels, the range of hypodense and hyperdense artifacts was determined, and the remaining tooth (non-affected tooth) was the complementary range (Fig. 3). The percentages of these areas were then calculated. When a hypodense artifact was very intense, it was impossible to select it in a black background, because the region had the same pixel value as the image background and the core region. In those cases, a white background image was used to assess the image artifact.

Fig. 3
Selection of the area corresponding to hypodense and hyperdense artifacts with the threshold tool. A. Hypodense artifact. B. Non-affected area of the tooth. C. Hyperdense artifact.

• Click for larger image

All analyses were performed using SPSS for Windows version 25.0 (IBM Corp., Armonk, NY, USA) with a significance level of 5%. Data were analyzed using descriptive and inferential statistics.

For the subjective analyses, hypodense halos, hypodense lines, and hyperdense lines were the dependent variables. These variables were scored according to the amount of artifacts as absence (0), moderate presence (1), or high presence (2). These ratings were statistically treated as continuous variables, where higher values indicate a higher amount of artifacts. For the objective analyses, the percentage of hypodense artifacts, percentage of tooth area, and percentage of hyperdense artifacts were the dependent variables. The variables were initially descriptively characterized using measures of central tendency, dispersion, and position, and the normality of their distribution was checked using the Kolmogorov-Smirnov test. The variables showed a non-normal distribution; thus, non-parametric tests were used.

The Mann-Whitney test was used to compare the artifacts between the alloys and between the cements. The Wilcoxon test was used to compare artifacts between exposure parameters and between the 2 devices.

In recent years, several studies have focused on assessing the possible variables that may influence the expression of artifacts on CBCT scans, aiming to better understand how to assess and improve image quality.4, 7, 10 Studies on this topic should always consider and respect the ALADAIP principle (“as low as diagnostically acceptable being indication-oriented and patient-specific”), using the most appropriate acquisition protocol for the specific purpose of the diagnosis.16

Metal posts were associated with greater artifact intensity and, therefore, compromised CBCT diagnostic performance in previous studies.4, 5, 15, 17 Although the use of metal posts has tended to decrease due to fiber posts’ aesthetic properties, their similar elastic modulus to dentin, and their tendency to induce favorable root fractures, metal posts are still used when greater retention is needed as a consequence of insufficient remaining dentin. According to Sarkis-Onofre et al.,18 fiberglass and metal posts presented good and similar clinical performance; therefore, choosing a particular metal alloy that presents lower artifact intensity on CBCT scans can improve the final CBCT image quality.19, 20

Metal alloys with higher atomic numbers attenuate more X-ray beams and produce higher artifact intensity. In this study, AgPd posts (atomic numbers 47 and 46, respectively) presented higher artifact intensity than nickel-chromium posts (atomic numbers 28 and 24, respectively). In accordance with this study, Panjnoush et al.19 found that cobalt-chromium alloys led to a higher presence of severe artifacts than titanium, and Chindasombatjareon et al.20 found that type IV gold alloys presented the largest artifact areas, followed by cobalt-chromium, titanium, and aluminum alloys.

Codari et al.7 also reinforced that high-density materials should be considered as being responsible for a variety of artifact manifestations. Amalgam, which is a heterogeneous material composed of high-atomic-number alloys (mercury [Z=80], silver [Z=47], tin [Z=50]), and a low-atomic-number alloy (zinc [Z=30]) presented higher artifact intensity than CuAl (Z=39 and 13, respectively) and titanium (Z=22), which led to substantial volume overestimation during image segmentation. Low-atomic-number alloy dental materials also presented less image noise.2

Endodontic post cements should present enough density to be distinguished from the surrounding structures. Root canal materials (obturation and sealing) produced a distinct degree of CBCT image artifacts given their density and composing elements.21 Few studies have assessed the capability of endodontic cements to induce CBCT artifacts. Freitas-e-Silva et al.22 found no difference between the evaluated endodontic cements in terms of their influence on vertical root fracture detection. Cavalcanti et al.23 found that BC Sealer induced significantly more imaging artifacts. In this study, the zinc-phosphate sealer presented higher hyperdense artifact intensity than the resin sealer for the CS9000 3D scanner; however, for the other assessed variables, no statistically significant differences were observed. This result may be explained by the small amount of cement used in the post casting and cementation technique, leaving only a small layer of cement, which may not have been enough to interfere with the overall artifact intensity. Therefore, other studies using different post sealers are needed to assess their effects on artifact intensity.

High-density materials, technical factors, and different CBCT devices should be considered as being responsible for a variety of artifact manifestations.7 The first step before acquiring a CBCT scan is to set the acquisition and exposure parameters based on the patients’ characteristics, indication for the examination, and area of interest. These parameters are paramount for acquiring a high-quality image following the ALADAIP principle.16, 19, 20, 24

In this study, higher tube current and lower tube voltage presented higher artifact intensity. The tube current is directly proportional to the amount of X-ray photons exiting the tube and, therefore, to the radiation dose, while tube voltage is related to the tube potential between the cathode and anode and determines the maximum X-ray energy.25 Low-energy X-ray beams have a greater probability of being absorbed, and increasing the mean energy of the beam causes beam-hardening artifacts. A higher peak kilovoltage reduces beam hardening, while changing the current and exposure time affects noise but not beam hardening.25

Lima et al.26 concluded that low-exposure protocols presented fewer image artifacts in CBCT scans, and Safi et al.27 also reported that lower mA parameters should be used for the detection of root fractures in teeth restored with intracanal posts. Lower current, higher tube voltage, and a smaller FOV are preferable in order to achieve fewer volumetric distortion artifacts in CBCT scanning.10 The results of the present study are in accordance with previous studies showing that higher tube voltage settings could reduce the artifact intensity of high-density materials.3, 10, 19, 28, 29

Different scanners present different FOV sizes, and the FOV size may be associated with a change in the scanning protocol, altering acquisition and exposure parameters.30 Voxel size may interfere with the detection of vertical root fractures and, in some scanners, is automatically associated with FOV size, with smaller voxels only possible for smaller-FOV protocols.31

Previous studies disagreed on the artifact intensity associated with the FOV size. According to Pauwels et al.,32 large-FOV scanners presented fewer artifacts than small-FOV scanners. However, Khosravifard et al.33 and Codari et al.7 reported significantly higher artifact generation by large- and medium-FOV scanners than by small-FOV scanners. In this study, the scanner with the highest FOV presented a lower amount of artifacts than the scanner with the lowest FOV. The inconsistent findings between studies may be related to the different scanners assessed, different acquisition and exposure parameters, and improvements in the performance of modern scanners. However, the use of a large FOV is not justified only for artifact reduction, because when all other parameters are similar, larger-FOV scanners result in higher X-ray doses.9, 33

In conclusion, high-atomic-number alloys, higher tube current, and lower tube voltage may increase the artifact intensity of CBCT scans. Operators must keep in mind the necessity of correctly choosing the acquisition and exposure parameters based on the ALADAIP principle.