Detection of vertical root fractures in the presence of artefacts by digital radiography and cone beam computed tomography

The correct diagnosis of vertical root fractures (VRF) in the presence of artifacts is a challenge for clinicians and endodontists. Moreover, there is controversy about which imaging technique is best for this purpose. In an in vitro model, we evaluated the diagnosis of VRF in teeth treated endodontically with and without intraradicular metal posts, using the Clark technique and cone beam computed tomography (CBCT), as well as the interference of artifacts with the diagnosis. Twenty-two first or second maxillary or mandibular premolars were included. Teeth were randomly allocated to three groups: G1 (two teeth without fracture with endodontic treatment and one with an intraradicular metal post); G2 (10 fractured teeth with endodontic treatment); and G3 (10 fractured teeth with endodontic treatment plus a metal post). The examiners recorded the presence or absence of fracture and its location and classified its type. There was no statistically significant difference between image acquisition systems. When differentiating the teeth (first vs. second premolars), there was a statistically significant difference among the examiners (p=0.020). However, when comparing the values obtained by the examiners regarding the visualization of the fracture site as well as the presence of fracture correlated with the presence of a metal post and angulation, there were no statistically significant differences Research, Society and Development, v. 10, n. 10, e284101018393, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i10.18393 2 (p>0.05). Digital radiographs and CBCT were similar for the diagnosis of VRF. High sensitivity was observed by CBCT image reconstructions. Therefore, the presence of metal posts generated artifacts, resulting in low sensitivity, specificity and accuracy.


Introduction
Vertical root fractures (VRF) may cause damage to mineralized tissues, periodontal ligaments and pulp, and has been the third most common reason for the extraction of an endodontically treated tooth (Andreasen et al, 2004). Excessive wear of dentinal walls for endodontic treatment and the use of metal posts may be associated with the cause of fractures (Khoshbin et al, 2018). Fractures are described as complete or incomplete, located in the root portion of the tooth and propagating coronarily, usually in the bucco-lingual direction (Mikrogeorgis et al, 2018). The first maxillary premolars treated endodontically are the second most affected group, representing 22.8% of all cases (PradeepKumar et al, 2016). The diagnosis is usually confirmed by clinical and radiographic characteristics, but not all the typical signs of a fractured root may be present in each case (Walton, 2017).
Radiographic images as a resource for the diagnosis of fracture are very important. However, radiographs may not confirm the presence of VRF. The fracture line is detected in only about 35.7% of cases (Wang et al, 2011). This is due to overlapping of root canals on the fracture line or other structures, the X-ray beam not parallel to the fracture plane, limited Research, Society andDevelopment, v. 10, n. 10, e284101018393, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i10.18393 3 sensitivity, and poor image quality (Edlund;Nair;Nair, 2011;Özer, 2011). To observe the fracture, the X-ray beam must pass perpendicular to the continuity solution, so that it is necessary to take radiographs with variations in horizontal and/or vertical angles (Andreasen et al, 2004). Cone beam computed tomography (CBCT) is an important tool for the diagnosis of VRF. Threedimensional images with reconstruction, manipulation of slices and absence of overlapping structures allow a more detailed assessment. Nevertheless, the presence of artifacts may hinder and/or modify the diagnosis, particularly when hypodense images are formed (Aristizabal-Elejalde et al, 2020).
Considering the challenge of interpreting the clinical diagnosis of VRF in the most affected group of teeth and aiming at its preservation, more modern imaging techniques can motivate clinicians and endodontists to seek their answers using CBCT, assuming its superiority for diagnostic accuracy (Talwar et al, 2016). In addition, there is controversy about which imaging technique is best for diagnosing VRF in the presence of metal posts. Based on this context, the purpose of the present in vitro study was to evaluate the diagnosis of VRF in endodontically treated teeth with and without intraradicular metal posts, using the Clark technique, CBCT, as well as the interference of artifacts with the diagnosis.

Study design and ethical issues
This in vitro study was approved by the Ethics Committee of the University of  following the guidelines of the Declaration of Helsinki. The study was performed in the Mechanical Testing Laboratory (Camaragibe, PE, Brazil), and the CBCT images were acquired at a private clinic of Oral and Maxillofacial Radiology (Brasília, DF, Brazil).

Sample selection and teeth assessment
Twenty-two uniradicular or birooted human teeth (first or second maxillary or mandibular premolars) freshly extracted for therapeutic reasons were inspected through stereoscopic glass (Callmex -Q705M, Florianópolis, SC, Brazil); ×10 magnification was used to determine whether the teeth had straight roots, a formed apex and absence of cracks and/or fractures.
The teeth were radiographed in the orthoradial and lateral directions to confirm the presence of a single and straight canal, the absence of calcifications, resorption, and/or anatomical variations.
The teeth were allocated to three groups: G1 (two teeth without fracture with endodontic treatment and one with an intraradicular metal post), G2 (10 fractured teeth with endodontic treatment), and G3 (10 fractured teeth with endodontic treatment plus a metal post). Endodontic preparations were performed by an endodontist using the Protaper Universal ® system, X-Smart ® engine and filled with F3 main cone and Sealer 26 cement (Dentsply Sirona, York, PA, USA).
For specimen preparation, the teeth were centered in 25.0×10.0 mm polyvinyl chloride rings and filled with chemically activated acrylic resin (JET ® , Artigos Odontológicos -Clássico, São Paulo, SP, Brazil). To simulate the periodontal ligament/artificial socket, a lead foil was used at the tooth/acrylic interface. Next, the lead was removed, and fluid silicone was added (Kit Express XT, 3M © São Paulo, SP, Brazil).
To perform the fracture, a force perpendicular to the long axis of the teeth was exerted using a digital spacer "D" (Dentsply Sirona, York, PA, USA) with 200 N compression, at a speed of 0.5 mm/min using the Kratos machine (IKCL3, Kratos Equipamentos Industriais, Cotia, SP, Brazil). The fracture was noticed by a sudden change in the graph and confirmed by inspection (Wilcox; Roskelley; Sutton, 1997).

Tooth exposure
The phantom consisted of four teeth inserted into fluid silicone (Kit Express XT, 3M © , São Paulo, SP, Brazil). The canine and first molar remained fixed, and premolars were assembled by simple random drawing without repetition. The Digora system was used for these procedures (Soredex, Orion Corporation, Helsinki, Finland). The fantom was positioned on an acrylic platform with a focus/phosphor plate at a distance of 30 cm and exposed to 60 kVp, 7 mA and 0.3 seconds, in a dental X-ray machine (Heliodent, Sirona Dental Systems Bensheim, Germany). Clark's technique was performed in the orthoradial position and at 25º on the right and left sides.
Tomographic images were captured with i-CAT New Generation ® equipment (Imaging Sciences International, Hatfield, PA, USA), with 120 kVp, 36.12 mAs and a 0.5 mm focal point. The acquisitions involved 14 bits of resolution and 0.20 mm of voxel size. The scanned volume was 13 cm in diameter and 6 cm in height. Axial, sagittal, and coronal images were reconstructed.
The volumetric acquisition was exported with a thickness of 0.2 mm in JPEG format. The images were captured by an oral radiologist.

Image acquisition
The images of the plaque system were coded and evaluated by three endodontists with more than 10 years of experience.
After the Kappa test, three examiners were selected. The software allowed manipulation of the image, brightness, and contrast.
The examiner received a questionnaire for evaluation of images selected at random, with the objective of recording the presence or absence of fracture, location of the fracture, and classification of the fracture (Aristizabal-Elejalde et al, 2020).
The CBCT images were coded and evaluated by three endodontic specialists. The examiners received guidance, images, and a questionnaire for image description. Tomographic slices were evaluated using the Microsoft Office Picture Manager image visualization program for Office 2010. The software permitted manipulation of image brightness and contrast according to the needs of the examiner. The tomographic evaluation was performed in two phases. The examiners first analyzed the selected image and then manipulated the slices along the three axes, evaluating the entire volume.
To assess the presence of artifacts, the images were saved in the XSTD (Xoran) format and exported to the Xoran/CAT program (Xoran Technologies, Ann Arbor, USA), permitting the execution of the axial, sagittal and coronal sections, and the assessment of total image volume. Super Mild Sharpen from the Xoran/CAT filter was used to adjust sharpness, brightness, contrast, and better viewing. The examiners performed analysis by answering the questionnaire about the presence and location of the fracture, the presence of artifacts and the influence of the artifact on the fracture diagnosis. The same computer was used, with a 23 "LCD" monitor model M2350D with Full HD resolution (Life's Good LG ® LG Corporation, Busan, South Korea) for the interpretation of the images of the plate system and CBCT.

Data analysis
The Kappa index was used to select the evaluators. The Chi-Square Test of the Statistical Package for the Social Sciences (SPSS) software, version 25.0 (SPSS Inc., Armonk, NY, USA) was used. The Fisher test was applied to determine the existence of an association between sensitivity, specificity, and accuracy. All tests were applied with 95% confidence.

Results
Kappa values among the evaluators for the analysis of digital radiographic images and CBCT in the two phases of evaluation are depicted in Table 1. In the first analysis, there was no statistically significant difference between image acquisition Research, Society and Development, v. 10, n. 10, e284101018393, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i10.18393 5 systems. In the second phase, there was a statistically significant difference between evaluators 1 and 3 (p=0.020) when differentiating between teeth (first or second premolar).   The data shown in Table 3 regarding the presence of a fracture and its correlation with the presence of a metal post and angulation demonstrate that there was no statistically significant association (p>0.05).

Discussion
The clinical signs of root fractures are nonspecific or common in endodontic and periodontal injuries. The diagnosis of VRF is a challenge for clinicians and endodontists and its early detection directly influences treatment planning (Tsesis et al, 2010). In the present study, there were no significant differences in sensitivity, specificity, or accuracy between the diagnostic methods based on the plate system and on CBCT scans, as also observed by others (Aristizabal-Elejalde et al, 2020). While the diagnostic methods were considered variable, both methods showed 50% sensitivity regarding the fracture. These findings agree with previous studies that reported specificity and accuracy values of 60% and 50% for radiography and CBCT scans (Chavda et al, 2014;Abdinian;Razavian;Jenabi, 2016). Nonetheless, a meta-analysis has detected the superiority of CBCT as an auxiliary tool for the diagnosis of VRF (Talwar et al, 2016).
Worldwide, periapical radiography is still the most widely used auxiliary method for the diagnosis of root fractures.
However, when there is no separation of the fragments, loss of substantivity or signs of periodontal involvement, the fracture line is often not visualized. Despite this, radiographic examination with variation of the horizontal angle should be encouraged as the first approach to the assessment of the presence of VRF. The images obtained with periapical radiographs, when associated with the Clark technique, as done in the present study, increase the fracture detection index (Andreasen et al, 2004;Chavda et al, 2014). On the other hand, CBCT allows three-dimensional reconstructions, eliminating overlaps and providing a more accurate analysis. The number of images is directly related to the amount of information needed to reconstruct the scanned object (Edlund;Nair;Nair, 2011). Accordingly, we observed a higher sensitivity value of 80% for volumetric analysis. However, former studies comparing the number of slices and their influence on the diagnostic capacity of VRF have observed that the number of images in different protocols did not influence the diagnosis (Bechara et al, 2013;Wanderley et al, 2017).
In the current study, two tomographic assessments were performed. In the first, only unmanipulated images were evaluated and in the second the examiners had access to the entire volume for interpretation and diagnosis of the teeth, which permitted the analysis of the sagittal, coronal, and axial views (Costa et al, 2012). To investigate the ability to diagnose root fractures by CBCT, some studies (Costa et al, 2012;Kamburoğlu et al, 2010) have used oral radiologists as examiners of the scanned images, a fact that explains the high kappa values for agreement between examiners. With the idea of obtaining data closer to clinical reality, we selected endodontists with experience in CBCT. For the analysis of unmanipulated images, the kappa values were moderate to substantial. For volumetric analysis, however, the kappa values disagreed, leading to the decision to select the examiner whose closest match was the correct diagnostic result. Notably, even using radiologist evaluators, some studies have attributed the compromised agreement between examiners to the difficulty in identifying the VRF depending on the tomograph used, the orientation of the fracture line, and the presence of intracanal materials (Costa et al, 2012;Kamburoğlu et al, 2010).
Although systematic reviews have reported that the sensitivity and specificity of CBCT are about 75% to 84% and 64% to 65%, respectively (Chang et al, 2016;Talwar et al, 2016), a more recent study detected 24% sensitivity when analysing unmanipulated radiographic images and 80% sensitivity for volumetric analysis, demonstrating a good ability to identify VRF (Aristizabal-Elejalde et al, 2020). Of note, we used a voxel of 0.2 mm, as recommended by other studies (Hassan et al, 2009;Kobayashi-Velasco et al, 2017;Wenzel et al, 2009). Wenzel et al. (2009) andÖzer (2011), when comparing voxel tomographs with 0.12 and 0.25 mm and 0.1 and 0.19 mm, respectively, did not detect statistical differences. Likewise, Wanderley et al. (2017), when using CBCT with 0.1 mm voxel and different scan times in endodontically treated teeth, obtained accuracy values of 90% to 93%, without statistical differences. In this respect, these studies reinforce the principles of ALARA, which advocates the use of protocols with lower doses of radiation.
The presence of biomaterials in the scanned area, such as intracanal materials, may influence the accuracy of the CBCT in the diagnosis of root fractures, compromising the quality of the image, reducing contrast, making it difficult or impossible to identify anatomical structures and making diagnosis difficult (Bechara et al, 2013). A former study evaluated the interference of four endodontic sealers with the diagnosis of VRF by means of CBCT. Sealer 26, due to its high radiopacity, revealed significantly more artifact formation (Brito-Júnior et al, 2014). In the present survey, the presence of gutta-percha, endodontic sealer, and a metal post interfered with the accuracy of the diagnostic method. In the analysis of unmanipulated images, the specificity was 74%, whereas low specificity values were observed in volumetric analysis, i.e., 20%, as also demonstrated elsewhere (Menezes et al, 2016). In this context, this evaluation allows a better visualization of structures, including beam hardening artifacts in general, which produce hyperdense lines in the image by optical illusion. Hypodense lines cause streaks that generate contrast, which can be confused with fracture lines, as observed in a similar manner by Edlund et al. (2011) and Wang et al. (2011). Interestingly, in our study, the examiner tended to give false-positive responses due to this occurrence, detecting fractures even when they were not present.
Herein, all teeth in the sample were treated endodontically. Considering the influence of the metal posts on the diagnosis of VRF, there was a reduction in accuracy, sensitivity, and specificity in the volumetric analysis. Nevertheless, Menezes et al.
(2016) when comparing the diagnosis of VRF by CBCT in teeth without endodontic material with gutta-percha and metal posts, obtained high accuracy, sensitivity, and specificity values. On the basis of the above considerations, we emphasize that clinicians and endodontists should be aware of the importance of exhausting digital radiographic techniques prior to requesting CBCT as a means of diagnosing VRF since radiographic images are easier to access and represent a simple diagnostic tool. On the other hand, when CBCT is necessary, its limitations should be considered regardless of the examiner's experience (Aristizabal-Elejalde et al, 2020). Based on the study design, the results should be interpreted with caution, since the number of examiners was limited for the analyses.

Conclusion
In summary, digital radiographs and CBCT provide similar findings in the diagnosis of VRF. High sensitivity was observed in CBCT image reconstructions. However, the presence of metal posts generated artefacts, resulting in low sensitivity, specificity, and accuracy of the imaging procedures used.