This experimental study was performed in five private radiology centers and maxillofacial radiology department of Hamadan Dental Faculty, Iran, which lasted for six months.

In this study, dry human mandible was used. The mandible was placed in a plastic container filled with water to simulate the soft tissue. Then, the mandible was scanned by means of five CBCT systems and one MSCT system [Table/Fig-1].

Image Assessment: Three oral and maxillofacial radiologists evaluated the scans under the same conditions (flat panel monitor; 1440×900 pixels resolution). In order to assess each system, certain software was used: Sygno, Fast Viewer for MSCT, on Demand 3D for Cranex 3D, Romexis Viewer for Promax, Sirona for Galileos and NNT Viewer for New Tom 3G.

They were asked to distinguish 10 anatomical landmarks namely the mental foramen, the mandibular canal, trabecular bone, cortical bone, dental pulp, dentin, enamel, incisive canal, PDL and lamina dura on the scans and then rank them from one to five based on the quality of the images using a five-point Likert scale [22]. The observers were allowed to observe the images in three planes of coronal, axial and sagittal or in each reconstructed plane. Moreover, they were allowed to adjust the contrast and brightness of the images. In order to prevent bias, each radiologist was blinded to the interpretations of others.

Data were analyzed using non-parametric Kruskal–Wallis test. Post-hoc comparisons of different systems were done by the Kendall’s W test. Also, the Friedman’s and Kendall’s W tests were used to assess the intra-observer agreement. Finally, in order to study the inter-observer agreement for the 10 landmarks, two observations were done in total and the results for each landmark were compared in two states. The Cramer’s V was used to compare the assesment by the three radiologists.

Ten landmarks in the mandibular bone were independently observed by three radiologists by means of six different software programs.

In order to rank the quality of images for observation of the landmarks, the five-point Likert scale was used ranging from one (undesirable observation) to five (excellent observation). Each radiologist assessed the landmarks twice. Since, the image quality for observation of the landmarks was reported using a five-point Likert scale, the Kruskal–Wallis and Mann–Whitney U tests (for comparison of average ratings), the Crammer’s V, the Kendall’s tau coefficient and the Friedman’s test were used for calculation of intra and inter observer agreements.

In order to study the differences in image quality among different systems, pairwise comparisons were made using the Mann–Whitney U test.

Finally, to assess the intra observer agreement in different sizes of Field of View (FOV), Wilcoxon signed rank and Friedman’s tests were used and the coefficient of agreement was calculated using the Cochran’s Q and the Kendall’s W tests.

In order to detect the general agreement coefficient of the three radiologists for the 10 landmarks, general reports by each radiologist at two different occasions for the six different systems were analyzed using the Cramer’s V.

For precise evaluation of the scans taken by the different CBCT systems (New Tom, Scanora, CRANEX 3D, Promax, and Galileos) and MSCT in our study, three oral and maxillofacial radiologists evaluated the 10 landmarks (mental foramen, mandibular canal, trabecular bone, cortical bone, dental pulp, dentin, enamel, incisive canal, PDL and lamina dura) twice in a dry human mandible scanned by these systems, and scored the image quality from one to five (excellent=1, good=2, average=3, bad=4 and very bad=5).

Significant differences were noted in the quality of observation of lamina dura (p=0.003), mental foramen (p=0.002), incisive canal (0.000) and PDL (p=0.002); but the difference was not significant for other landmarks [Table/Fig-2].

Among six CBCT systems in two observations (in total), CRANEX acquired the highest score in six landmarks and Promax ranked second in three landmarks. Scanora and Galileos acquired the highest score in one landmark and the MSCT and New Tom acquired the lowest score among all [Table/Fig-3].

Pair wise comparisons of different systems showed that image quality of some systems had significant difference [Table/Fig-4].

[Table/Fig-5] explained the amount of the intraobserver agreement for different sizes of FOV for each system.

There was a weak correlation in the radiologists’ interpretation for lamina dura (Cramer’s V=0.293), which was not significant (p=0.62). Similarly, there was a weak correlation for mental foramen (Cramer’s V=0.216) and this correlation was not significant either (p=0.22). The three radiologists had complete agreement in observing the mandibular canal. The three radiologists had complete agreement (100%) in observing the cortical bone as well. There was a strong correlation in the radiologists’ interpretation of dental pulp (Cramer’s V=0.675), which was significant (p=0.01). There was a strong correlation in the radiologists’ interpretation of dentin (Cramer’s V=0.737), which was significant (p=0.003). There was a weak correlation in the radiologists’ interpretation for the incisive canal (Cramer’s V=0.899), which was not significant (p=0.178). There was a strong correlation in the radiologists’ interpretation for the enamel (Cramer’s V=0.737), which was significant (p=0.003). There was a weak correlation in the radiologists’ interpretation for the PDL (Cramer’s V=0.221), which was significant (p=0.204). Also, there was an average correlation in the radiologists’ interpretation for the trabecular bone (Cramer’s V=0.537), which was not significant (p=0.01).

As a whole, it can be said that the difference was significant for four landmarks namely mental foramen, lamial dura, incisive canal and PDL [Table/Fig-6].

The use of three-dimensional imaging systems has improved dento-maxillofacial diagnosis [26]. Optimal image quality of vital structures has many benefits for patient and clinician especially in implant therapy.

In the present study, CRANEX 3D had the best image quality in the assessment of 10 landmarks and then stood Promax, Scanora, Galileos and New Tom 3G respectively. MSCT had the lowest image quality. Since New Tom and MSCT have higher kVp than other systems [Table/Fig-1], it can be concluded that high kV and, in turn, higher scattered radiation in these devices result in higher noise of the images which makes the visualization of fine structures difficult. In practice, the choice of CBCT system depends on whatever you need from imaging. In major maxillofacial surgery or orthodontic treatment plan we need Large Field Of View (FOV), so New Tom has the selection criteria but in other treatment planning such as implant therapy the CBCT systems with lower kVp and small FOV seem to be a better option for detecting fine details in vital structures and bone status.

In a study by Liang X et al., it was found that the image quality of CBCT was comparable or even higher than that of MSCT and there were some differences among CBCT systems [27]; in the current study, we concluded that the image quality of CBCT was higher than that of MSCT.

Liang X et al., claimed that bone trabecula and Periodontal Ligament (PDL) were less observable and variable among different systems [27]. In our study, the incisive canal had the poorest quality of observation and PDL did not have a good quality of observation either. This difference may stem from the differences in CBCT systems, which were used in the current study and the study by Liang X et al., as well as the method of image quality evaluation.

Kamburoglu K et al., [28] used landmarks for evaluation of image quality of CBCT and reported different results from the present study. However, regarding the same landmarks (trabecular bone and mandibular canal), both studies concluded that the image quality of these two landmarks in different systems was significantly different.

The study by Hashimoto K et al., was similar to our study and they reported results similar to ours, indicative of higher accuracy of CBCT (3DX) than that of MSCT [22].

Loubele M et al., showed that image quality was better significantly in CBCT than MSCT with respect to lamina dura and PDL observation [29]. They used a dry maxilla for image quality evaluation. Although the systems and the jaw, used in their study were different from the present study but the results were similar.

Lofthag HS et al., evaluated image quality of Accuitomo CBCT in three FOV [30]. Seven radiologists assessed the images. Intra-observer agreement was good and interobserver agreement was moderate. In the present study, interobserver agreement in three landmarks wasn’t good.

Lamina dura was given the least score by the first radiologist and there was a weak agreement among the three radiologists with regard to this landmark as well as mental foramen.

Interestingly, with regard to the mandibular canal and cortical bone, all three radiologists completely agreed (100%) that the image quality of these landmarks was good. The first and third observers gave a low score to the incisive canal, which indicates low image quality of this landmark.

However, limitation of this study was the invitro nature of the study and also we didn’t assess image quality of maxilla. Further studies should be carried out for in-vivo evaluation of image quality with other CBCT systems.

The present study concluded that CBCT systems had better image quality and lower noise than MSCT. Among CBCT systems selection of the best device depends on the target landmark that is going to be evaluated.