Accuracy Analysis of Extraoral 3D Scanning in the Development of Dental Prosthetic

Abstract

Objective

The study has evaluated the accuracy (trueness and precision) of seven extraoral scanners when scanning two different types of jaws: simplified jaw with sharp edges and abutments and realistic jaw with natural teeth. The accuracies of extraoral scanners were compared, and their compliance with the required clinical accuracy levels was discussed.

Material and methods

Ten scans were made with each scanner for both models. The comparison of the selected dental scanners relied on reference scans made for both models. Trueness, precision, and the distribution and value of laboratory scan points’ deviations were assessed for each scanner across the models.

Results

The trueness for the model of the simplified jaw with abutments ranged from 16.15 to 49.78 μm. The measured precision values for the same model ranged from 4.33 to 29.49 μm. For the model of the realistic jaw with natural teeth, the trueness results ranged from 11.32 to 24.55 μm, while the obtained precision values were between 2.29 and 18.06 μm.

Conclusion

The revealed dissimilarities in the accuracies of scanners and their ranking when scanning different models lead to the conclusion that model selection is critical for the research design. All the scanners met the clinical accuracy requirements and are suitable for use in laboratories for scanning jaws with abutments and jaws with natural teeth. However, the accuracy values reported by the manufacturers of scanners are better than those obtained in this study. Furthermore, the results suggested that blue light scanners outperform white light and laser scanners.

Introduction

The introduction of digital technologies has greatly impacted the dental industry in recent decades (1–4). CAD/CAM systems have been introduced in a variety of dental offices, laboratories, and production centers due to the emergence of new materials suitable for industrial production (1, 2, 5, 6). These systems enabled a reduction of the production time, an increase in productivity (6), and an improvement in dimensional reliability. Therefore, the introduction of the CAD/CAM systems has enabled the time- and cost-effective production of individual prosthetic replacements (3).

CAD/CAM systems consist of three basic components: 3D scanner, computer software, and production technology (3). The 3D scanner converts the geometry of a prepared tooth or an entire jaw into a digital shape, based on which a prosthetic replacement is designed. Furthermore, the software derives a CAD model of a prosthetic replacement according to the geometry of the obtained digital shape. Finally, the designed CAD model is used to prepare the code for production (CAM process).

3D scanners for dental use are classified as intraoral and extraoral. Scanning represents the first step in designing and producing individual prosthetic replacements (7). Thus, scanner accuracy is a vital prerequisite for the final product quality. If the model's surface is scanned with insufficient accuracy, the obtained product will have wrong measurements and shape, thus causing cement loosening (8, 9) and potentially leading to periodontal diseases (10, 11), secondary caries (12), or microleakage (13).

Numerous studies evaluated the accuracy of 3D scanners for dental use (14–27). In these studies, the accuracy of scanners is commonly assessed with regards to trueness and precision. Trueness is defined as the deviation of the object's scanned dimensions from its actual dimensions, whereas precision is regarded as either the deviation of dimensions between multiple scans of a particular object (21) or as the deviation of dimensions within an individual scan (22). High trueness signals that the scanner produces the scans very close to the actual dimensions of the scanned object. High precision indicates that the scanner gives reproducible scan results.

Previous studies utilized various validation methods and metrics for trueness and precision, hampering comparisons across different studies. Furthermore, the existing studies assessed the accuracy of scanners using models of different shapes and materials. For example, scanners achieved better accuracy when scanning a single tooth or prepared tooth models than when the entire jaw model was scanned (17).

Extraoral scanners generally have better accuracy than intraoral scanners, which varies less along the dental arch (28). In addition, the accuracy of intraoral scanners is inadequate for scanning edentulous arches with implants (29). These distinctions stem from differences in scanning technologies. Intraoral scanners capture images or videos while moving along the arc of the dental arch (30), as opposed to extraoral scanners, which capture the entire dental arch positioned at different angles. Individual images are then combined to obtain a virtual 3D model (31). The software recognizes common points in the images, that is, two coordinates of each point (x and y) are estimated in the figure. In contrast, the z-coordinate is determined independently by calculating the object's distance from the cameras. The z-coordinate estimation results in errors that further propagate through the process.

Sharp edges are another factor that may have a detrimental effect on the scanning quality. Previous studies highlighted the differences in the accuracy of scanners when scanning smooth surfaces in comparison to scanning sharp edges or undermined areas (14). The main reason for such discrepancies is that 3D scanners sample 3D shapes' surfaces uniformly, without aligning the samples with the sharp edges and corners of the original shape. The interpolating triangle meshes thus chamfer the sharp features, resulting in significant errors (32).

The main objective of this study is to evaluate the trueness and precision of seven extraoral scanners when scanning two different types of jaws: simplified jaw with abutments and sharp edges and genuine jaw with natural teeth. The accuracies of extraoral scanners were compared, and their compliance with the required clinical accuracy levels was discussed.

Material and methods

Experimental procedure

The existing studies typically analyze scans of only one model (33). Nevertheless, their results indicate the influence of geometry complexity on 3D scanning accuracy. In this study, two models were utilized to enable comparison between scanners regarding model types. The first model, A model, was an arch-shaped model mimicking the mandibular arch with abutments designed using the CAD software solution SolidWorks. The model comprised six identical abutments (9 mm in height, 8 mm in diameter at the bottom, and with a 6° inclination), arranged on a flat surface in the positions of the right second molar, right second premolar, right lateral incisor, left lateral incisor, left second premolar, and left second molar (Fig. 1). B model was created in Exocad by scanning a maxillary arch with healthy natural teeth from the right second molar to the left second molar (Fig. 1). Both designs were milled on Dental Plus RS5 milling machine from a polyetheretherketone (PEEK) disk. This material was selected due to good mechanical and chemical resistance and optical properties that enable a good scan of the surface.

Figure 1.

Models A (left) and B (right) utilized in this study

Seven extraoral scanners listed in Table 1 were evaluated in this study. Ten scans were made with each scanner for both models. The scan data were exported in standard tessellation language (STL).

Table 1. Characteristics of scanners used in this study.

	Scanner
	UP360	UP300	DWS3	D810	D2000	Freedom HD	Identica
Manufacturer	Up3D	Up3D	Dental Wings	3Shape	3Shape	DOF	Medit
Light	Blue light	Blue light	Laser	Red laser	Blue LED Multi-line	White light LED	Blue LED
Cameras	2 x 1,3 MP	2 x 2 MP	2	2 x 5 MP	4 x 5 MP	2 x 2 MP	1
Accuracy (according to manufacturer)	< 6 μm	< 10 μm	< 20 μm	< 15 μm	< 8 µm	< 10 µm	< 10 µm

The comparison of the selected dental scanners relies on reference scans made for both master models using the industrial 3D scanner Atos Core 135. Before scanning, the reference scanner was calibrated according to VDI/VDIE 2634 Part 3 (16) with a calibration panel (GOM Type CPI40-170-40288). The calibration identified 0.001 mm probing error (sigma), -0.005 mm probing error (size), 0.005 mm sphere spacing error, and 0.006 mm length measurement error.

3D analysis

Each scanner's trueness was assessed using deviation values yielded by comparing the reference model scans with the corresponding ten laboratory scans. Scans were compared using 3D analysis based on the ICP algorithm using Geomagic Control X – software for 3D quality control and dimensional inspection (16, 17, 22). First, each laboratory scan was aligned with the reference scan using the best fit alignment with a 100% sampling ratio, 100 iteration count, and maximal average deviation of 0.001 mm. Then, the 3D deviation (x, y, and z) from the reference scan point was calculated for every laboratory scan point. Finally, since the root mean square (RMS) captures the magnitude of deviations from the reference scan, it was calculated for each analysis (N=140) and used to determine the scanners' trueness (15–17).

Precision was assessed by superimposing each (except the first) scan of a particular scanner over the scanner's first scan. Standard deviation across all points was calculated for each superimposed pair (N=126).

Finally, the software provided color-coded deviation maps, thus showing the distribution and value of laboratory scan points' deviations from the reference scans.

Statistical analysis

The statistical analysis was performed in R, and a significance level was set to 0.05. The Shapiro-Wilk test was utilized to assess the distributions' normality, whereas the Levine's test was used for testing the homogeneity of variance. If the tests indicated no violation of the assumptions, ANOVA was used for the general significance of the differences, followed by the Tukey test to review significant differences between scanner pairs. However, in the case of heteroscedastic data, Welch ANOVA was utilized, and the post hoc Games-Howell test was performed. When the normality assumption was violated, the Kruskal-Wallis test was used, followed by the Mann-Whitney-Wilcoxon test with Holm correction for multiple comparisons.

Results

The experiments assessed the scanners' accuracy by measuring their trueness and precision.

Trueness

Trueness is defined as the deviations of the scanned models' points from the reference model. It is measured using RMS values (15–17). For each scanner and both models, ten scans were analyzed.

A model

The scanners' RMS values when scanning Model A are presented in Table 2.

Table 2. RMS values (μm) for scanners when scanning A model.

RMS
	Mean	SD	Median	Min	Max
Model A
UP360	37.07	2.73	37.85	30.28	39.73
UP300	30.94	1.61	30.45	28.92	34.38
DWS3	49.78	5.25	51.03	38.34	56.37
D810	24.09	0.31	24.06	23.62	24.63
D2000	18.58	2.34	18.67	13.76	21.92
DOF	29.63	1.28	29.78	26.22	30.98
Identica	16.15	0.6	15.94	15.56	17.44

Identica has the lowest mean RMS (16.15 μm), and its RMS values span the smallest range between 10 scans (from 15.56 to 17.44 μm). D2000 achieved the second-best result (18.58 μm), but with a broader range (from 13.76 to 21.92 μm), thus indicating that D2000 is a less consistent scanner. DWS3 demonstrated the highest mean RMS (49.78 μm), and its RMS values vary the most (from 38.34 to 56.37 μm).

The UP360 data violate the normality assumption (p = 0.0217), and the deviation from the homoscedasticity assumption was detected (p = 0.0006). The Welch-ANOVA test showed statistically significant differences between the tested scanners (p = 2.2∙10^-16). Significant differences were found between all scanner pairs except between DOF and UP300, and Identica and D2000 (Table 3).

Table 3. Statistical significance (p-values) of differences in the scanners' RMS values on A model.

	UP360	UP300	DWS3	D810	D2000	DOF
UP300	3.5∙10-4
DWS3	1.64∙10-4	6.35∙10-6
D810	1.26∙10-6	2.36∙10-6	1.16∙10-6
D2000	8.91∙10-11	5.07∙10-9	8.44∙10-9	4.51∙10-4
DOF	5.09∙10-5	0.442	4.68∙10-6	1.58∙10-6	5.22∙10-8
Identica	4.93∙10-9	5.02∙10-11	8.35∙10-8	2.46∙10-12	0.094	6.9∙10-12

B model

The scanner's RMS values when scanning Model B are presented in Table 4.

Table 4. RMS values (μm) for scanners when scanning B model.

RMS
	Mean	SD	Median	Min	Max
Model B
UP360	15.15	0.21	15.09	14.81	15.53
UP300	19.24	0.53	19.27	18.23	20.21
DWS3	24.55	0.88	24.91	23.23	25.55
D810	20.9	0.29	20.93	20.44	21.44
D2000	16.12	0.61	16.12	15.3	16.97
DOF	23.19	1.19	23.33	21.66	24.87
Identica	11.32	0.19	11.27	11.2	11.87

The Identica scans again had the lowest mean RMS (11.32 μm) and the smallest RMS range (11.2 – 11.87 μm). UP360 achieved a better result when scanning Model B than Model A and is ranked second with the mean RMS of 15.15 μm and a lower range (14.81 – 15.53 μm). D2000 also showed a low mean RMS value (16.12 μm), and its RMS varied less on Model B than on Model A (15.3 – 16.97 μm). DWS3 is again ranked last, as it has the highest mean RMS (24.55 μm). DOF's mean RMS equals 23.19 μm.

In this case, Identica's data violate the normality assumption (p = 1.67∙10^-5), and a violation of variances' homogeneity assumption was detected (p = 6.61∙10^-8). Statistically significant differences between the tested scanners were found for all scanner pairs, except between DOF and DWS3 (Table 5).

Table 5. Statistical significance (p-values) of differences in the scanners' RMS values on B model.

	UP360	UP300	DWS3	D810	D2000	DOF
UP300	8.17∙10-10
DWS3	2.6∙10-10	1.26∙10-9
D810	1.65∙10-13	9.85∙10-6	1.33∙10- 6
D2000	0.0080	1.02∙10-8	7.73∙10-13	2.36∙10-10
DOF	3.14∙10- 8	6.21∙10-6	0.1130	0.0020	3.38∙10-9
Identica	4.42∙10-14	0	5.72∙10-11	0	1.39∙10-9	3.38∙10-9

For both models, the results are presented graphically in Fig. 2.

Figure 2.

The scanners' trueness comparison - Box-plot diagrams of RMS (μm) values for both models

Precision

Following the works by Su and Sun (18) and Kim et al. (21), this study conceptualized a scanner's precision as the deviation between its scans. More precisely, for each scanner, its first scan was compared to every other scan via superimposition and the assessment of deviations. Overall, 18 scan comparisons were made for each scanner (nine for each model).

A model

The scanners' standard deviations when scanning A model are presented in Table 6.

Table 6. Standard deviations of scanners (μm) when scanning A model.

Standard deviation
	Mean	SD	Median	Min	Max
Model A
UP360	10.38	2.04	9.75	8.54	15.39
UP300	19.20	5.66	16.98	13.35	30.35
DWS3	29.49	2.33	28.59	27.15	33.55
D810	6.56	0.61	6.77	5.37	7.21
D2000	9.61	2.25	9.56	6.37	13.21
DOF	4.33	0.81	4.23	3.05	6.15
Identica	9.44	3.99	7.07	6.20	18.42

On average, DOF showed the best precision performance, with the mean standard deviation between its scans equaling 4.33 μm, followed by D810, whose mean standard deviation between scans was 6.56 μm. Further, D810 scans showed the smallest variation in standard deviation results (5.37 – 7.21 μm). In contrast, DWS3 had the weakest average performance (mean standard deviation = 29.49 μm), whereas UP300's results spanned across the broadest range (from 13.35 μm to 30.35 μm).

The data collected using Identica, UP360, UP300, and DWS3 scanners violate the normality assumption (p = 0.017, p = 0.0048, p = 0.0462, and p = 0.0055, respectively). No significant violation of the homoscedasticity assumption was found (p = 0.0844). No statistically significant difference was found among Identica, UP360, and D2000 scanner pairs (p = 1). Similarly, Identica and D810 did not differ significantly (p = 0.37565). Nevertheless, all other pairs differed significantly (Table 7).

Table 7. Statistical significance (p-values) of differences between scanners regarding their precision on A model A.

	UP360	UP300	DWS3	D810	D2000	DOF
UP300	0.00148
DWS3	0.00086	0.01111
D810	0.00086	0.00086	0.00086
D2000	1	0.00086	0.00086	0.01111
DOF	0.00086	0.00086	0.00086	0.00148	0.00086
Identica	1.00000	0.00864	0.00086	0.37565	1	0.00086

B model

Similar to A model, the scanners' standard deviations when scanning B model are presented in Table 8.

Table 8. Standard deviations (μm) for scanners when scanning B model.

Standard deviation
	Mean	SD	Median	Min	Max
Model B
UP360	3.61	0.12	3.59	3.37	3.82
UP300	3.57	0.40	3.46	3.08	4.17
DWS3	18.06	1.07	17.96	16.38	19.84
D810	7.18	0.53	7.25	6.00	7.91
D2000	7.37	1.46	7.70	4.99	9.20
DOF	3.63	0.38	3.69	3.00	4.02
Identica	2.29	0.05	2.29	2.17	2.36

In the case of B model, Identica outperforms other scanners, as the superimpositions of its scans yielded the lowest mean standard deviation (2.29 μm), as well the smallest range of standard deviation values (2.17 – 2.36 μm) followed by UP300, with the mean standard deviation equal to 3.57 μm and the standard deviations ranging from 3.08 to 4.17 μm. The worst results were obtained using DWS3, whose superimpositions again had the highest mean standard deviation (18.06 μm).

No violation of the normality assumption was detected for any of the scanners. However, a violation of the homoscedasticity assumption was found (p = 1.783∙10^-11). Thus, the Welch ANOVA test was employed, thus revealing statistically significant differences between the tested scanners (p = 2.2∙10^-16). More precisely, significant differences were found for all scanner pairs, except the differences between UP360-UP300, UP360-DOF, DOF-UP360, and D2000-D810 (Table 9).

Table 9. Statistical significance (p-values) of differences between scanners regarding their precision on B model.

	UP360	UP300	DWS3	D810	D2000	DOF
UP300	1
DWS3	3.43∙10-9	3.22∙10-11
D810	3.22∙10-11	1.2∙10-9	2.76∙10-8
D2000	6.59∙10-4	4.2∙10-4	4.59∙10-10	1
DOF	1	1	1.38∙10-10	1.9∙10-9	5.08∙10-4
Identica	2.94∙10-11	1.3∙10-4	1.86∙10-9	2.76∙10-8	7.64∙10-5	4.83∙10-5

Figure 3 depicts the precision results for both models.

Figure 3.

The scanner's precision comparison - Box-plot diagram of standard deviations (μm) for both models

Deviation distribution analysis

Color-coded maps are utilized to depict the distribution of scan data deviations from the reference data. The analysis enables identifying critical areas, i.e., the areas where the scans have the highest deviation from the reference. For each scanner, a laboratory scan whose RMS value was the closest to the corresponding scanners' mean, RMS was selected as a representative scan and utilized in the deviation analysis of color-coded difference images (Table 10).

Table 10. Selected scans for the deviation analysis.

	Model A		Model B
Scanner	Scan No.	RMS value (μm)	Scan No.	RMS value (μm)
UP360	7	36.7	5	15.12
UP300	10	30.94	5	19.25
DWS3	8	50.53	7	24.95
D810	10	24.1	2	20.91
D2000	5	18.43	5	15.98
DOF	6	29.63	6	22.62
Identica	7	16.3	1	11.32

A model

Figure 4 shows the color-coded maps for all scanners when scanning A model. The deviations are colored as follows. Areas with nominal deviations, ranging from -10 to 10 μm, are marked in green following Emir and Ayyildiz (16). Yellow and orange tones show areas with positive deviations, ranging from 10 to 120 μm, modeled based on Medina-Sotomayor et al. (22) and in accordance with clinical accuracy limits (8). Areas with deviations higher than 120 μm are colored red. In contrast, blue tones highlight areas of negative deviations, ranging from -10 to -100 μm. Areas with deviation values higher than -120 μm are colored dark blue.

Figure 4.

Color-coded maps for Model A: a) UP360; b) UP300 c) DWS3 d) D810 e) D2000 f) DOF g) Identica

Figure 4 shows that lateral areas of the abutments and the side of the arch along the model have predominantly positive deviations (a, b, c, f, and g) or slightly negative deviations (d and e). The surface where abutments are located has minimal deviations – nominal (a, b, d, e, and g) or slightly negative (c and f).

B model

Figure 5 presents the color-coded maps of all scanners when scanning B model. Again, the deviations are colored following the same rules as described in the previous section.

Figure 5.

Color-coded maps for Model B: a) UP360; b) UP300 c) DWS3 d) D810 e) D2000 f) DOF g) Identica

The color-coded maps look more uniform in this case than in the case of A model. Nevertheless, one can note a superior performance of Identica since map g has few small areas out of the nominal range. Other maps show negative deviations on incisal edges and occlusal surfaces (a, b, c, d, and e), whereas positive deviations are seen on palatal surfaces (a, b, c, d) and fissures (d, e, f).

Discussion

The existing studies define and measure trueness and precision differently. Researchers expressed a scanner's trueness using the mean RMS values (15–17, 27) or the mean MAD values (14, 21–23, 25). The precision of a particular scanner is commonly measured as the mean variability within its scans (i.e., variability in one scan’s deviations from the reference) by, for example, averaging the standard deviations of each scan's data (14–16, 18, 22, 25). Others conceptualized precision as the variability among scans of a particular scanner. In such a case, precision was assessed by superimposing the scans, calculating the absolute deviations, and finally averaging them to obtain a precision value (18, 21, 24, 26). Compared to intraoral studies, the accuracy of extraoral scanners was found to be higher and it varied less along the dental arch (17, 18, 21, 22).

In contrast to most existing studies, this study utilized two models to evaluate the scanners' trueness and precision. When scanning A model, Identica showed the highest trueness (i.e., the lowest RMS). However, its performance did not differ significantly from D2000, whose RMS value was only slightly higher. Based on trueness performance, the two scanners were followed by D810, then DOF and UP300 (which are statistically indistinguishable), UP360, and, finally, DWS3. Identica also achieved the best trueness result when scanning Model B, but other scanners were ranked differently: UP360, D2000, UP300, D810, DOF, and DWS3.

When measuring precision on A model, the best scanner was DOF, followed by D810. Identica, UP360, and D2000 shared third place, UP300 was penultimate, and DWS3 was again ranked last. However, Identica was the most precise scanner when scanning B model UP300, UP360, followed by DOF, D810, D2000, and DWS3. Therefore, except for the precision score on A model, Identica achieved the best results. In contrast, DWS3 was consistently ranked last. The variations in scanners' ranks when scanning different models highlight the impact of model selection on research findings.

The trueness results for A model can be contrasted to the results obtained by Emir and Ayyildiz (16) since the study investigated extraoral scanners' performance using a similar model of a simplified edentulous jaw with abutments. The trueness values (as proxied by RMS) ranged from 17.4 to 33.3 μm. The study presented herein obtained slightly different values (16.15 – 49.78 μm). Furthermore, in the study by Emir and Ayyildiz (16), scanner D2000 achieved the highest trueness result (17.4 μm). However, when testing this scanner within the study at hand, it scored slightly worse mean trueness (18.58 μm) and was ranked second.

For B model, the trueness results range from 11.32 to 24.55 μm, and the obtained precision values are between 2.29 and 18.06 μm. When scanning a similar model, Park et al. (17) found that trueness of extraoral scanners ranges from 19.6 to 67.3 μm. In addition, the extraoral scanner achieved the 8.67 – 24.33 μm precision on a model similar to Model B, as reported by Su and Sun (18). Another related study found that extraoral scanners achieved trueness from 7.7 to 28.6 μm and precision from 4 to 19.5 μm when scanning a single abutment (15). Furthermore, de Villaumbrosia et al. (14) reported that results of scanning the prepared tooth ranged from 29 to 46 μm regarding MAD (i.e., trueness), and from 37.6 to 50.6 μm for standard deviation (i.e., precision).

Table 1 reveals that the accuracy values reported by the scanner manufacturers are not compatible with the presented results. The reason might lie in different scanning conditions and the utilized accuracy parameters. Hence, the detected discrepancies highlight the necessity of reporting detailed information regarding the scanned model, as well as the way scans are performed and evaluated.

However, there is no established limit to the value of clinical accuracy of dental scanners, as current literature reported the required accuracy being below 50-75 μm (15), 80 μm (34), 100 μm (35), 120 μm (8), or 150 μm (36). Most studies follow McLean and Fraunhofer (37), who argue that the fit on the margin between the batter and the replacement must be less than 120 μm. It should also be noted that the CAD/CAM process error should be calculated with respect to clinical accuracy limit value, rendering the scanner deviation error smaller.

For both models and at the level of the entire jaw, all tested scanners achieved values of trueness and precision that meet the minimum specified limit (50 – 75 μm). Nevertheless, it is insufficient to reflect on clinical accuracy using only the accuracy values aggregated for whole arches (i.e., jaws). The accuracy on the sides of the abutments and teeth shown in the color-coded maps should be compared. The color-coded maps show a substantial difference in deviations on the surface when scanning A model. Deviations are especially large at the abutment areas where several scanners have reached results close to maximum allowable positive deviation. Since these areas are vital for the final function of the prosthetic replacement, the deviations should remain limited. Large deviations at the abutment areas might occur due to simplified design with completely flat, smooth jaw surfaces and equal, steep, and smooth abutment surfaces. These high and steep abutments can create shadows that prevent properly capturing the shaded areas unless the scanner tilts the model plate sufficiently. Such significant deviations can occur due to the inadequate rotation angle of the scanner model plate (for structured light scanners) or the scanner camera (for laser scanners). As in previous studies, scanners had problems when scanning sharp edges or undermined areas (14). Such results are obtained because 3D scanners sample surfaces of 3D shapes uniformly and do not attempt to align the samples with the sharp edges and corners of the original shape. This problem can be solved by adding algorithms for sharp features recovering in triangulations (29).

With regards to scanning technology, blue light scanners (Identica, D2000, UP360, and UP300) achieved generally better results than white light scanners (DOF) and laser scanners (D810, DWS3). These findings correspond to the ones reported by Emir and Ayyildiz (16), showing that the structured blue light scanners produce fewer errors, and their outputs have greater repeatability.

To further explore the relationship between the differences in extraoral scanners' accuracy and the utilized scanning technologies, the presented study should be expanded to include a larger number of scanners of different generations. Furthermore, the research can be improved by utilizing additional models, such as models of whole arches with prepared teeth, models with scanning bodies, or models of an individual prepared tooth. Finally, another step in gaining a detailed insight into the accuracy of all types of dental scanners in different situations in dentistry would be to contrast the results obtained by extraoral scanners to those obtained by intraoral scanners.

Conclusion

The presented in-vitro study demonstrated that the scanning accuracy depends on the type of the scanned model. The revealed dissimilarities in the scanners' accuracy and their ranking when scanning different models lead to conclusions that model selection is critical for the research design, as it can significantly impact the obtained findings. Therefore, there is a need for creating a scanning methodology which will ensure the appropriate evaluation of scanners accuracy and enable comparison across the results of different studies.

Of the tested scanners, Identica showed the best accuracy when scanning jaws with abutments and jaws with natural teeth. On the other hand, DWS3 obtained the lowest accuracy results. Furthermore, the accuracy values reported by the scanner' manufacturers are better than those obtained in this study. Nevertheless, all scanners satisfied the clinical accuracy requirements and are suitable for use in laboratories for scanning jaws with abutments and jaws with natural teeth. With regards to scanning technology, the obtained results demonstrated that blue light scanners (Identica, D2000, UP360, and UP300) generally outperform white light scanners (DOF) and laser scanners (D810, DWS3).

Data Availability

All data are available upon request.