Biomechanical complications and accumulation of bacterial biofilm are the main causative factors in the development of peri-implantitis and peri-implant bone defects. Fenestrations and dehiscences can affect the success of implant therapy and can subsequently result in progressive bone loss and implant failure. As a result, early detection of these defects is important for preserving the implants with the aid of radiographic examinations[7, 28]. According to several studies [3, 9, 10, 29], when peri-implant dehiscence and fenestration are suspected, periapical and panoramic radiographs cannot be used for the evaluation of interproximal bone level around dental implants, and CBCT should be employed when peri-implant dehiscence or fenestration is suspected.
Bovine rib bone blocks were used in the current study because it was believed that the bone density and proportions between the cancellous and cortical bone in bovine ribs were similar to those in the mandible of humans. This was consistent with de-Azevedo‐Vaz et al.[2],Saberi et al.[9], and Schwindling et al.[30], who used bovine rib models to evaluate peri-implant bone defects. For the selection of acquisition parameters, the smallest FOV of the machine was chosen as it was recommended by Pinheiro et al.[31] to decrease the artifacts from the surrounding tissues and optimize the radiation dose. Based on Vasconcelos et al.[32] findings, voxel size didn’t have an effect on beam hardening artifacts production next to titanium implants, so we chose a voxel size of 0.08 to enhance image quality and improve spatial resolution. For tube voltage, we choose 90 kVp for all radiographic protocols as it was suggested by Pauwels et al.[33], who reported that 90 kVp gives the optimum image quality and less noise than images obtained with 75 kVp.
For optimization of radiation dose, several previous studies[34–36] concluded that decreasing tube current in the presence of metallic objects may increase the magnitude of artifacts but has no significant effect on the diagnostic ability of CBCT images. Fontenele et al.[34] showed that decreasing the tube current had no effect on the detection accuracy of vertical root fractures in the endodontically treated teeth next to zirconium implants. Also, Sawicki et al.[35] found that the difference in tube current didn’t affect the assessment of peri-implant bone level. Based on these findings, we chose 12 mA (the highest mA offered by the CBCT machine used) for HD-CBCT and 2 mA (the least mA offered by the CBCT machine used) for LD-CBCT.
Low-dose CBCT protocols are associated with a significant decrease in radiation dose in comparison to high-dose protocols. For the current study, the dose area products of the protocols used were 48.7 mGycm2 for LD-CBCT and 396.92 mGycm2 for HD-CBCT. Despite the representation of the dose in the form of DAP, it still indicates that the use of the low-dose protocol significantly decreases the dose when compared with the high-definition protocol. Although the reduction in radiation dose of LD-CBCT protocols is significant, an insufficient amount of research is available for its application in implant assessment. A study conducted by Liljeholm et al.[37] showed that ultra-low dose protocols of CBCT can be used for pre-implant radiographic assessment. Another study held by Cardarelli et al.[38] stated that using a low-dose protocol with 180 degrees rotation angle has a significant effect on decreasing metal artifacts around the implants and can be used to assess peri-implant bone level. However, for detection of peri-implant dehiscence and fenestrations, de-Azevedo‐Vaz et al.[5] recommended using the full scan protocol (360 degrees) for the detection of peri-implant dehiscence.
The results of our study showed that there was no statistically significant difference in fenestration and dehiscence detection between HD-CBCT and LD-CBCT when the MAR tool was not applied. Our study results support a previous study conducted by Schwindling et al.[30] to compare the accuracy of HD-CBCT and LD-CBCT in the detection and classification of peri-implant bone lesions, which stated that there was no significant difference in the diagnostic accuracy of both protocols. Also, in a study conducted by Schriber et al.[39], no difference was found between low-dose and high-dose protocols in the detection of buccal peri-implant dehiscence defects. Our results were in line with the results of Aktuna-Belgin et al.[40], who evaluated the efficiency of two different CBCT doses (low dose and ultra-low dose) in the detection of peri-implant fenestration and dehiscence defects, and they found that the diagnostic accuracy was not affected in the two protocols used.
The incorporation of MAR algorithms into CBCT units by manufacturers has resulted in a steady growth in the utilization of these approaches. Different previous studies evaluated the effects of the MAR algorithm on the artifacts of CBCT images, and they stated that it can reduce the standard deviations of grey value and increase image quality by increasing contrast to noise ratio (CNR)[41–44]. The application of these strategies has been examined, but the findings have been uneven. Fontenele et al.[45] found that the MAR tool has a negative effect on the diagnostic accuracy of vertical root fracture in the presence of intracanal filling. Also, Kamburoglu et al.[46] evaluated the effects of four MAR protocols (off, low, medium, and high) for the assessment of periodontal and peri-implant defects, and they found no difference in diagnostic accuracy with any MAR protocol. However, Bagis et al.[47] recommended the use of the MAR tool for the detection of peri-implant fenestration defects.
According to the results of the current study, using the MAR tool decreased AUC, sensitivity, accuracy, and NPV values when applied to both HD-CBCT and LD-CBCT. Only specificity and PPV were the same whether MAR tool was applied or not. It means that the detection of true defect blocks was significantly decreased by the application of MAR tool, while the no defect intact blocks could be detected correctly with and without the application of MAR tool with no difference. This was consistent with the findings of both De-Azevedo-Vaz et al.[2] and Sheikhi et al.[28]. De-Azevedo-Vaz et al.[2] stated that the MAR algorithm didn’t improve the diagnostic accuracy of fenestration and dehiscences. Sheikhi et al.[23] found that sensitivity and accuracy values for fenestration and dehiscence defects were higher when the MAR algorithm was absent, but specificity was equal when MAR was present and absent, which was almost similar to our findings.
In the current study, when MAR tool was applied, the AUC values and the diagnostic values revealed that peri-implant fenestrations were more correctly diagnosed than peri-implant dehiscences. Similar findings have been reported by de-Azevedo-Vaz et al.[2, 5], Sheikhi et al.[23], and Salemi et al.[26], and they explained these findings by claiming that as dehiscence has just an inferior border, it is more difficult to identify than fenestration, which has both superior and inferior borders.
MAR algorithms employ several methods to minimize metal artifacts, such as iterative reconstruction methods, projection-correction methods, and reconstruction-correction methods. The majority of these methods consider metal artifacts as missing data[48, 49]. The technique of the MAR algorithm of the CBCT machine used in the study was not explained by the manufacturer. Unfortunately, the disadvantages of these techniques include the elimination of all attenuation data from high-density objects. When using the MAR algorithm, there is always lost data that cannot be reconstructed, which may cause modification in the image and can affect the diagnostic accuracy[42, 50]. Also, according to Fontenele and Mancini et al.[51, 52], the impact of the MAR algorithm becomes more prominent when the image artifacts are increased. Hence, it explains the decrease in the diagnostic ability of LD-CBCT compared with HD-CBCT when the MAR tool was applied to both in the current study, as we changed the tube current from 12 mA in HD-CBCT to 2 mA in LD-CBCT. Thus, the production of artifacts from the implants was increased[52, 53].
The in vitro design is one of the limitations of our study. In clinical situations, patient movement artifacts may affect the quality of the final image and affect the diagnostic accuracy, but this is the only ethically appropriate protocol to assess the recommended parameters without subjecting patients to unneeded CBCT scans[19]. Also, the presence of metal restorations, teeth, and surrounding tissues in the clinical situations may make the detection of these defects more difficult. Another limitation is that peri-implant dehiscence and fenestration defects were prepared using a diamond bur with a definite border and differed from the naturally occurring defects, which have tapered borders and are more difficult to detect[54].