Most orthodontic patients have crowding [1]. Therefore, the primary goal is often to expand the alveolar arch [2]. To eliminate crowding and create space, the teeth must be moved vestibularly.

Adverse side effects of orthodontic treatment include mucosal defects and bone lesions. In addition, the literature shows a relationship between alveolar bone resorption and orthodontic therapy [3,4]. However, bony dehiscences and fenestrations also occur in the untreated dentition [5]. Therefore, it is essential to adequately assess the bony conditions of the jaw before treatment. Accordingly, the survey of the peridental bone supply is an important pretherapeutic measure [3,6]. Alveolar bone morphology is a limiting factor for orthodontic treatment, making analysis of peridental bone before orthodontic treatment inevitable [5,7].

Bone defects are increasingly identified through two-dimensional imaging techniques. The fact that information from a three-dimensional space is compressed into a two-dimensional image may result in distorted information [8]. Information may be lost on a panoramic radiograph because it is “superimposed” and thus cannot be faithfully represented [8,9]. Compared with conventional imaging techniques, CBCT has the great advantage that many more bony defects can be detected and measured [10,11,12].

Since the bone morphology can be depicted faithfully in CBCT images, the maximum tooth movement during therapy is easier for the practitioner to recognize. CBCT diagnosis in orthodontics helps the practitioner to develop an individualized treatment plan for the patient [4,6,7,13]. Although three-dimensional imaging can depict bony structures more faithfully than two-dimensional imaging, CBCT is not necessarily used pretherapeutically in daily practice due to the higher radiation exposure.

CBCT standard protocols have a very high effective dose of 220 μSv. Over the last five years, there has been an increase in study protocols that show a significant reduction in effective doses by reducing mAs values. The reduction of mAs values results in a linear reduction in effective radiation [6,14,15,16,17]. In the literature, a reduction of the effective dose by up to -79% can be found [16]. These values place 20-40 μSv effective dose per cone beam exposure in the range of two-dimensional radiographs.

Some authors recommend different radiation protection concepts, such as ALARA (As Low As Reasonably Achievable) [18,19] to ALADA (As Low As Diagnostically Acceptable) [20,21]. The difference here is that in the ALADA concept, the radiation dose is kept to a maximum of what is diagnostically necessary. A razor-sharp image is not necessary; only the diagnostically relevant part must be adequately displayed.

In addition to reducing the effective dose, diagnostically sufficient image quality is also of great importance. Some studies have addressed whether the reduction of mA values is accompanied by a deterioration of image quality and whether the image is then still diagnostically sufficient [6,23]. A sound knowledge of the change in effective dose as a function of image quality plays an important role in the implementation of the latest concept ALADA-iP (acceptable, indication-based, and patient-specific) [11].

With this concept and the adjustment of the appropriate parameters, it is possible to produce diagnostic-quality CBCT images while significantly reducing the radiation dose. However, a reduction in mAs values may lead to a degradation of image quality. Therefore, it needs to be clarified whether this deterioration in image quality has a negative impact on diagnosis or whether the images still have sufficient diagnostic value. In some studies, a reduction to 4 mA was performed, which still resulted in a diagnostically adequate image. Reducing mA values to 4 mA did not show a significant impact on image quality [6,23].

A recent meta-review examined publications addressing the radiation dose of CBCT images in dentistry [24]. The meta-review calls for further studies on effective dose directly related to image quality. Therefore, the present study follows this call and investigates the direct relationship between mAs reduction and image quality. The aim is to be able to perform optimal bone measurements in a CBCT while achieving an immense reduction in effective radiation.

In the present study, the CBCT images of 14 macerated human skulls were examined. For this purpose, 318 teeth with 636 dehiscences (oral and vestibular) were included in the study. Two CBCT images were acquired from each skull using the ProMax 3D Mid CBCT unit (Planmeca, Helsinki, Finland). Images were acquired using two different protocols, resulting in a total of 28 data sets examined (protocol A: field of view (FoV) 20 x 17 cm, voxel size 200 μm, current voltage 90 kV, exposure time 12 s, current intensity 6.0 mA, current intensity per second 72 mAs; protocol B: field of view (FoV) 20 x 17 cm, voxel size 200 μm, current voltage 90 kV, exposure time 12 s, current intensity 3.2 mA, current intensity per second 38.4 mAs). Both protocols are reduced in their mAs values compared to standard protocols, whereas protocol B shows a stronger mAs reduction than protocol A.

The skulls were additionally scanned with the Trios 3 intraoral scanner (Shape, Copenhagen, Denmark, see Table 1). The digitally generated three-dimensional image matched the original object very well due to the precise scanning and is therefore considered the gold standard in the present work.

Subsequently, the scans were imported into Nemotec software (Nemo Studio 2019, Nemotec, Madrid, Spain) with the CBCT NemStudio datasets and measured. NemoStudio is a dental software used in wide areas of dentistry, such as implantology and orthodontics. In the software, it is possible to merge three-dimensional radiographs (such as CBCT with STL datasets) by superimposition (matching). In the present work, this aspect was of great importance. The measurement points on the X-ray image as well as in the STL file from the intraoral scan (gold standard) were thus at the same position. Thus, measurement inaccuracies could be reduced.

The DICOM and STL data sets were uploaded individually to the Nemo Studio software. In the software, the teeth were aligned and overlaid (matched) in all three spatial planes. After matching, the vestibular and oral bone dehiscences were measured individually.

First, the height of the enamel-cement interface was marked in the DICOM dataset. This mark was also seen in the STL dataset at the height of the enamel-cement interface.

The distance between the enamel-cement interface and the upper margin of the crestal alveolar bone was measured as dehiscence. The measurement was performed on both the vestibular and oral sides in both the STL and DICOM data sets.

For each protocol, four measurement series per dehiscence were measured by one rater. A total of 5088 measurements were available.

A single-factor analysis of variance was performed, and the measurement time points were compared to statistically determine reproducibility. The Intraoral Scanner (STL) measurements represent the gold standard in this work. A correlation analysis was performed in person to determine the correlation between the measurements from the DICOM datasets for protocols A and B with the measurements from the STL datasets. A paired-samples t-test was also performed for direct comparison of the two protocols, and Bland-Altman plots were generated. Data was collected in Excel spreadsheets and transferred to IBM SPSS Statistics for statistical analysis.

Reproducibility of bone dehiscences in intraoral scan, protocols A and B

Intraoral scan

The reliability of the intraoral scan showed no statistically significant differences in its four measurement series, neither for oral, nor for vestibular. Single factor analysis of variance was performed to compare the four measurement time points. According to the f-statistic, the values of the four measurement series have not statistically significantly different from each other (F (3.1206) = .013; p = .998; F (3.1206) = .034; p = .991).

Figure 5 shows the differences in the measured values of the dehiscences of the intraoral scanner together with their frequencies. A normal distribution with a rightward shift can be seen. Here, the absolute mean difference is present as a maximum value at 0.10 mm.

Protocol A

The reliability of the measurements of protocol A showed no statistically significant differences in its four measurement series, neither for oral, nor for vestibular. Single factor analysis of variance was performed to compare the four measurement time points Single factor analysis of variance was performed to compare the four measurement time points. According to the f statistic, the values of the four measurement series have no statistically significant difference (F (3.1207) = .017; p = .997; F (3.1207) = .040; p = .989).

Figure 6 represents the deviations in the measured values of the dehiscences for Protocol A, together with its frequencies. For the diagram of the reliability of Protocol A, a slight shift to the right of up to 0.2 - 0.3 mm can be seen. Most of the deviations are in the range of -0.1 mm to 0.1 mm. The largest mean deviation is 0.016 mm (Table 5).

Protocol B

The reliability of the measurements of protocol A showed no statistically significant differences in its four measurement series, neither for oral, nor for vestibular. Single factor analysis of variance was performed to compare the four measurement time points. The values of the four measurement series have no statistically significant difference (F (3.1183) = .009; p = .999; F (3.1183) = .003; p = 1.000). For the measured values of the dehiscences of Protocol B, the Mean (M), Standard Deviation (SD) and mean difference were also determined using statistical calculations (Table 7). Figure 7 presents the deviations of the measured values of the dehiscences for Protocol B, together with its frequencies. A shift in the positive direction can be seen with respect to the bar chart of the reliability for Protocol B (Figure 4). In the negative range there are deviations from -0.1 mm to 0 mm, and in the positive range the deviations extend up to 0.4 mm. The largest mean deviation is 0.027 mm (Table 7).

Deviation of bone measurements of both protocols from the gold standard

Protocol A

There is no statistic significant difference between the vestibular measurements from the CBCT dataset for Protocol A and the measurements from the Intraoral Scan (STL). Pearson’s correlation coefficient is 0.960 for the vestibular measurements, representing a very high correlation. A t-test confirmed this strong correlation (P < 0.01). The values of the vestibular dehiscences of the DICOM dataset for protocol A were plotted, along with the respective values of the STL dataset, in a Bland-Altman plot. The mean of both measurements was plotted on the x-axis, and the difference of both measurements was plotted on the y-axis. The LoA (Limits of Agreement) are defined as +1.0 mm for the upper limit and -0.81 mm for the lower limit. Within these limits, 95% of the differences and mean values of the measurements are located. Figure 8 shows a very high density of values on the straight line of the mean difference. This means that most of the values have very small to no measurement differences. Most of the values are very close together to, just below, on, or above the straight line mean difference.

A few of the values are well outside the mean difference, and represent outlier values at 1%. The mean absolute difference between the DICOM protocol A and STL measurements is 0.09 mm for the vestibular measurements, well below the clinical tolerance range.

Protocol B

There is no significant difference between the oral measurements from the CBCT data set for Protocol B and the measurements from the Intraoral Scan (STL). The Pearson correlation coefficient for the oral measurements is 0.981, representing a very strong correlation. A t-test confirmed this strong correlation (p > 0.001).

The values of the oral dehiscences of the DICOM dataset for Protocol B were plotted with the respective values of the STL dataset in a Bland-Altman diagram. The mean of both measurements was plotted on the x-axis and the difference between both measurements was plotted on the y-axis. The LoA (Limits of Agreement) are defined as +0.57 mm for the upper limit and -0.46 mm for the lower limit. Within these limits, 95% of the differences and mean values of the measurements are located. Figure 10 shows a very high density of values on the straight line of the mean difference. This means that most of the values have very small to no measurement differences. Most of the values are very close together. A few values are far outside the mean difference, and represent outlier values of 0.2%.

The mean absolute difference between the DICOM protocol B and STL measurements is 0.05 mm for the oral measurements, well below the clinical tolerance range.

Deviation of bone measurements between protocols A and B

The determined vestibular dehiscences in mm are almost identical in protocols A and B. There are no significant differences between the measurements of vestibular dehiscences for both protocols (p-value = 0.509). The mean values of the vestibular dehiscences of both protocols are therefore very close to each other, and show a very high level of agreement. Figure 12 shows the mean values of the vestibular dehiscences of protocols A and B in a bar chart. The mean value of the vestibular dehiscences for Protocol A is 3.24 mm. The mean value of vestibular dehiscences for Protocol B is 3.27 mm. There is a difference of 0.03 mm in the mean values for both protocols.

The determined oral dehiscences in mm are almost identical in protocols A and B. There are no significant differences between the measurements of oral dehiscences for both protocols (p-value = 0.515). Chart 11 presents the mean values of oral dehiscences for protocols A and B in a bar graph. The mean value of oral dehiscences for Protocol A is 2.84 mm. The mean value of oral dehiscences for Protocol B is 2.88 mm. There is a difference of 0.04 mm in the mean values for both protocols. There is a small difference of 0.04 mm in the mean values of protocols A and B. The mean values for the oral dehiscences of both protocols are therefore very close to each other, showing a very high level of agreement.

Due to the high prevalence of bone defects, bone measurements should be taken by X-ray prior to orthodontic treatment in order to establish an adequate treatment plan. CBCT has the advantage of showing dehiscences in oral and vestibular directions more naturally. Since most patients in orthodontics are children and adolescents, the issue of reducing the effective dose is particularly important here.

There are currently over 85 different CBCT devices on the market, all with different features and functions [10]. In particular, it is difficult to evaluate the effective doses between them because the parameters are set differently. Various analyses have shown that the effective dose of different CBCT devices ranges from 5 μSv to 1073 μSv [25, 26,27].

These studies show the wide range of effective doses of the various CBCT devices on the market. According to them, the different doses depend on the device, the FoV, and the setting parameters such as current voltage, tube intensity, and exposure time [52]. This range is highly unsuitable for clinical use. Various investigators have attempted to reduce the radiation dose by changing the parameters and using different protocols. Studies have been performed in which the Fov, current voltage kV, voxel size, or exposure time were changed [10,27,28,29,30]. In the studies, some investigators note a reduction in radiation exposure, but this is often accompanied by a degradation in image quality. The change in image quality due to noise or a decrease in image contrast represents the limits of altered parameters.

In our study, the radiation dose with the manufacturer’s recommended protocols is 263 μSv at 360 mAs. The effective dose with the manufacturer’s lower protocols is 117.2 mAs at 83 μSv. For comparison, a panoramic slice exposure is approximately 20 μSv.

The optimal CBCT protocol is the one that has the lowest possible radiation exposure without compromising diagnostic value through image quality degradation. In our case, this means reducing the mAs enough to keep the radiation exposure at the level of a standard panoramic radiograph.

The purpose of this study is to compare the radiographs to the gold standard and determine if further reduction of mAs has an impact on the accuracy of diagnosis of peridental bone.

As also our present study Several studies have investigated how to modify the parameters of CBCT protocols to achieve a reduction in radiation dose while maintaining the same image quality or adequate diagnostic accuracy. Some of these studies have shown that image quality was adequate for the diagnosis of bone defects when mAs were reduced to a maximum of 4 mA. Similar to the present study no significant differences were found between the different protocols.

Sur et al. 2010 investigated the reduction of mAs from 6 mAs to 2 mAs in macerated skulls. Ten investigators compared the results and concluded that the reduction in mAs had no effect on diagnostic image quality for objective bone measurements. There was sufficient agreement for qualitative assessment of CBCT images.

Tangari-Meira et al. (2017) performed studies with different mA values between 4 and 10 mA and found no significant differences in image quality. In 2018, Panmekiate et al. analyzed the optimal balance between parameters of CBCT images. They found that reducing the mA value by up to 40% is possible to reduce the radiation dose while maintaining adequate image quality.

A significant reduction in effective dose can be achieved by decreasing the mAs value without significantly decreasing image quality when assessing bone ratio [6]. The present work follows up on these findings by evaluating two different mA-reduced protocols for their accuracy in diagnosing bone defects. The results of this work show that the diagnostic value even for the mA-reduced protocol (3.2 mA) for linear bone measurements with the previously mentioned parameters has no significant difference compared to CBCT images with higher mA values.

Thus, the results of our study are consistent with those cited in the finding that reducing mAs down to 4 mA does not significantly degrade image quality.

Furthermore, the reproducibility of the measurement method depends on the conditions of the measurement method. In the present work, measurements were performed according to a stan-dardized measurement methodology based on published studies by Fuhrmann (1996), Evangelista (2010), Lee (2012), Jäger (2015), and Elefant (2015). Here, the alignment of the respective teeth to be measured in all three spatial planes was a condition to ensure reproducibility.

With this precise alignment, it is possible to steadily place the teeth in the same position and measure them. Determining the enamel-cement interface in the STL file of the intraoral scan proved to be a challenge. The measurement of the lower anterior teeth was particularly critical due to the smooth transition from enamel to root dentin. Since the transition from enamel to root cementum was often difficult to determine in the scans on the lower anterior teeth, the scatter of mean deviations tended to be highest here compared to the other tooth groups. The single factor analysis of variance confirmed the reproducibility of the bone dehiscences in the present intraoral scan.

With regard to the high clinical prevalence and the low effective doses, evaluation of the peridental bone availability seems to be a main justified indication prior to orthodontic treatment planning. The introduced mAs-reduced ultra-low-dose protocols provide the same or even less effective dose when compared with the effective dose of COR. This means, the clinician has 3D treatment planning information with less dose than with traditional COR.

In the future, the clinical discussion shouldn´t be directed to the decision panoramic x-ray/lateral ceph or CBCT, but rather which specific CBCT protocol should be used for orthodontic treatment planning.

MAs-reduced ultra-low-dose CBCT protocols with a resolution of 200 μm allow for a reliable quantification of peridental bone structures, especially bony dehiscences. These results are important for daily planning of orthodontic tooth movements and its limitations. With the investigated mAs-reduced ultra-low-dose CBCT protocols the effective dose can be - especially for children - significantly reduced to values lower than with 2D images.