Metal artifacts produced by dental implants influence the image quality of anatomic structures and complicate bone assessment for adjacent implants. Many studies have investigated artifacts in CBCT, but there are issues in quantifying artifacts based on image quality [19]. The qualitative assessment, based on the observer judgement, is helpful in evaluating the decreased effects of artifacts and diagnosis, but this assessment cannot compare the efficiency of systems and cannot be used in quality control.
It’s difficult to provide an objective and quantitative assessment of artifacts in CBCT [20]. There are no standard parameters to determine the quantity of artifact effects on voxel [21]. Studies have evaluated artifacts by various parameters, such as MGV and SD of gray values, and CNR [22]. The MGV approximates the radiopacity and radiolucency produced by metals. Higher SD represents more noise and lower image quality, while CNR shows effect of artifacts on image contrast and helps compare CBCT units. A higher value for CNR shows better image contrast and quality [23]. Although SD and CNR provide valuable information, they require precise interpretation, because voxel sizes and intrinsic noises also affect image quality [24]. Furthermore, the gray value of a CBCT image depends on various factors such as exposure settings, the patient’s position, the device model, and the ROI’s area or size. This may have an impact on the results and impair their precision. According to recent studies, to reduce the undesirable effects of inherent noise, the amount of streak artifacts decreased by MAR was automatically counted in MATLAB using canny edge detection, which is an approach that is less dependent on the image’s gray value [20, 25].
Our results show that object sizes representing dimensions of patient soft tissue had significant effect on artifacts, such that objects equal to FOV and 20% smaller than FOV had more homogenous MGV, higher CNR and lower SD, resulting in less intense artifacts and better image quality. S phantoms produced better quality images than M phantoms, but differences were insignificant. Objects larger than the FOV had more intense artifacts. It has been proved that exomass, or objects outside the FOV that remain between the focal spot and the receptor, influenced image quality. Artifacts related with exomass directly influence voxel value and image noise [26]. To decrease artifacts, it is better to apply larger FOV in patients with larger size. It is essential to mention that increasing FOV size may decrease image resolution. Our findings are in agreement with results reported by Seet et al. who showed that increased phantom size influences CT value, while decreasing precision and image quality [14]. Previous studies have reported that beam scatter has a direct relation to object size, such that increased object size augments beam diffraction [27]. This result was confirmed in objects larger than FOV, however, in XS phantom, higher SD and lower CNR were detected compared to S and M sized objects. So, it can be assumed that decreased soft tissue volume doesn’t necessarily increase image quality. The existence of more artifacts in XS objects compared to S and M objects could be related to the fact that beam streaks does not cross over the borders of objects. Therefore, an artifact concentration could be expected in smaller sized objects [28]. Spread of beam streaks occurs in larger objects, further improving image quality around the metallic material. This issue could be applied to larger objects as long as it does not cause an exomass artifact by overextending the selected FOV. To prevent truncating the axial section data, manufacturers recommend using a FOV equal to the patient’s maximum dimensions. However, imaging with FOV equal to object dimensions also produces exomass artifacts. Very small parts of medium phantom were positioned in exomass on closer inspection. This shows CBCT sensitivity to exomass artifacts. A difference of 20 mm [2] in axial section is enough for producing artifacts resulting from exomass [14].
The results showed that ROI location also influences image quality. The highest SD was observed in areas of the trabecular bone that were intrinsically inhomogeneous. Knowing that artifacts cause appearance of regions with lower gray values near the implant has helped in preventing the diagnosis of false positive diagnoses of peri implantitis [29]. It was reported that trabecular microstructure parameters observed in micro-CT used for quantification of artifacts are more appropriate than SD of the gray values to evaluate artifacts in bones [22]. In our study, ROIs near the implant produced more SD. In other words, artifacts were higher in locations close to metal objects. Similar to our findings, previous studies have shown that distance has a significant effect on the production of artifacts, and increased distance decreases artifacts [11, 30, 31]. Mancini et al. investigated three ROI location (1.5, 2.5, and 3.5 cm) and demonstrated non-significant difference for the SD in distances of 1.5 and 2.5 cm, while the SD was lower in 3.5 cm location [32]. Fontenele et al. did not observe significant differences for artifacts and image quality in locations of 1.5, 2.5, and 3.5 cm [33].
The CBCT units had significant effects on the image quality. Although the main purpose of this study was to assess the effect of object size on the amounts of produced artifacts, we could also make a comparison between the two CBCT systems. However, the multi-faceted nature of these systems, such as hardening, Voxel size, voltage, the FOV and current intensity can affect the obtained results [34]. NewTom VGI unit were characterized by less SD and more CNR, therefore less artifacts were observed in NewTom VGI unit than HDXWill CBCT unit, which may be related to the higher kVp value of the Newtom VGI CBCT scanner. Furthermore, NewTom scanner has an automatic exposure system via “safebeam sensor” which will adapt itself to the patient’s factors. Probably, this is why by modifying mA in a NewTom device the image quality difference between phantoms of different sizes is much less than in a HDXWill device. The fact that mA, as an interfering variant, could not be fixed was a limitation of the present study. Codari et al. also showed that the NewTom VGI system has better diagnostic power than the Picasso Trio system due to its 360 degree rotation and increased data receiving ability [35]. Moreover, Kamburoğlu et al. showed lower beam hardening artifacts production in the NewTom VGI system than in the Promax system and suggested the use of NewTom VGI systems for patients with multiple restorations, prosthesis, and implants [29]. The differences in diagnostic value between different CBCT units are mainly attributed to their resolutions [36].
In our study, dental implants were placed in sheep mandible bone blocks to simulate in vivo conditions. We used ballistic gelatin to simulate soft tissue. Other studies have used water, ice, wax, ultrasound gel, and acrylic resin to simulate soft tissue [17, 37, 38]. A recently study has introduced ballistic gelatin as the best simulator of soft tissue [39]. Despite the inhomogeneous nature of human tissue, we used homogenous materials for evaluating the artifacts. Homogeneous materials have advantages such as allowing for more precision in the estimation of gray values. Most studies conducted on CBCT imaging are done in vitro due to ethical considerations. Because of FOV size and various exposure parameters, studies done on patient’s image data fail to provide accurate information for specific variables. In vitro studies, like this study, allow for precision in controlling interfering variables. Future studies can investigate voxel size, exposure parameters, high-density materials in other positions in dental arch, and an increased number of implants.
This study is an in vitro study and the data must be cautiously used under clinical condition because X-ray interferences are different in each patient [30]. In this study, the dental implant was positioned in central region of the FOV which may produce lower levels of noise compared to marginally positioned metal object [40]. This study has not examined the motion artifacts that can cause problems in clinical conditions. In addition, the exact effects of artifacts on trabecular structures were not investigated due to the intrinsic limitations of CBCT systems.