Graphical abstract
Abstract
Objective
To evaluate a novel methodology using industrial scanners as a reference, and assess in vivo accuracy of 3 intraoral scanners (IOS) and conventional impressions. Further, to evaluate IOS precision in vivo.
Methods
Four reference-bodies were bonded to the buccal surfaces of upper premolars and incisors in five subjects. After three reference-scans, ATOS Core 80 (ATOS), subjects were scanned three times with three IOS systems: 3M True Definition (3M), CEREC Omnicam (OMNI) and Trios 3 (TRIOS). One conventional impression (IMPR) was taken, 3M Impregum Penta Soft, and poured models were digitized with laboratory scanner 3shape D1000 (D1000).
Best-fit alignment of reference-bodies and 3D Compare Analysis was performed. Precision of ATOS and D1000 was assessed for quantitative evaluation and comparison. Accuracy of IOS and IMPR were analyzed using ATOS as reference. Precision of IOS was evaluated through intra-system comparison.
Results
Precision of ATOS reference scanner (mean 0.6 μm) and D1000 (mean 0.5 μm) was high. Pairwise multiple comparisons of reference-bodies located in different tooth positions displayed a statistically significant difference of accuracy between two scanner-groups: 3M and TRIOS, over OMNI ( p value range 0.0001 to 0.0006). IMPR did not show any statistically significant difference to IOS. However, deviations of IOS and IMPR were within a similar magnitude. No statistical difference was found for IOS precision.
Conclusion
The methodology can be used for assessing accuracy of IOS and IMPR in vivo in up to five units bilaterally from midline. 3M and TRIOS had a higher accuracy than OMNI. IMPR overlapped both groups.
Clinical significance
Intraoral scanners can be used as a replacement for conventional impressions when restoring up to ten units without extended edentulous spans.
1
Introduction
CAD/CAM was introduced in dentistry for single-unit restorations over thirty years ago, and advancements in technology has made it possible to produce complex multi-unit restorations on teeth and implants [ ]. An essential part of the workflow was the indirect digitization process by laboratory scanners of gypsum models poured from traditional analogue impressions [ ]. Parallel to this technology, CEREC, a commercialized intraoral scanner (IOS), made it possible to digitize the dental status in situ [ ]. However, the early CEREC IOS were limited to single-tooth restorations and came as an integral part in a proprietary workflow accompanied by an in-house milling machine.
Since the late 2000’s, there has been a rapid increase in the number of commercial IOS with scanners capable of capturing full dental arches [ ]. Yet, with an open platform allowing for third-party CAD/CAM manufacturing, and rapid growth in the number of IOS, there is limited data to validate if IOS can replace conventional impressions.
Varying terminology exists in the science of metrology for explaining intra- and inter-system variations. Although some research groups in dental literature have used the ISO 5725 [ ], this study has adopted the more common definition of accuracy and precision. Hence, accuracy being defined as the ability of a measurement to match the actual value, and precision defined as the ability of a measurement to be consistently reproduced.
Several in vitro studies have evaluated IOS applying methodology of 3D Compare Analysis based on varying software best-fit alignment, ranging from single-units to full dental arches [ ]. Some studies comparing IOS and conventional impressions have found IOS to demonstrate a statistically significant lower accuracy [ ]. However, variations in study design, execution and material properties pose a challenge in comparing results between existing studies, even when conducted under optimal conditions and far from clinical reality. A study by Muller et al. [ ] found different scanning strategies amongst IOS to result in statistically significant differences in accuracy. Gimenez et al. [ ] found differences due to bias when comparing scans from experienced and inexperienced users scanning implant models.
Material properties and scanner technology can further affect IOS accuracy. Nedelcu and Persson [ ] found sizeable differences in non-coating scanners and Li et al. [ ] have shown that increased translucency in scanned objects resulted in lower accuracy. Yet several studies evaluating IOS in vitro have used a variety of materials, from gypsum [ ], and polyurethane [ ], acting as almost perfect diffusers, to metal [ ], having a higher level of specular reflection. These material properties differ greatly from naturally translucent teeth, and may either oversimplify the clinical reality of non-coating scanners, or contrarily put the non-coating scanners to great disadvantage when scanning metal.
Models with translucency and refraction index close to enamel and dentin [ ], have been adopted by Renne et al. [ ] in vivo. Yet, there are other properties beyond translucency, such as opacity, iridescence, surface gloss, and fluorescence which vary individually between human teeth, as does the effect of tooth age, thickness and color of the enamel and underlying dentin [ ]. Any in vitro model will further be limited compared to clinical reality as the models will not take into account non-attached tissues, such the sulcus, lips, floor of the mouth or tongue. IOS use pattern recognition to stitch multiple images with partial overlapping areas into full models [ ], and the scans will, voluntarily or not, include a certain amount of data of non-attached tissues. Any movement between measurements in the underlying surfaces may cause improper stitching of the scan. These are elements that are practically impossible to model in vitro.
There are few in vivo studies of IOS. Ender et al. [ ], focused on the precision of numerous intraoral scanners and conventional impressions in quadrant and complete-arch impressions. The studies do not take into consideration the accuracy of each system.
A novel approach to evaluating accuracy of IOS and conventional impressions in vivo was conducted by Kuhr et al. [ ], utilizing a reference-system based on bonded spheres onto teeth and a custom transfer guide. However, as the accuracy analysis is only made at the location of the spheres, there is limited possibility in evaluating accuracy of any surface deviations beyond the spheres, or to assess the in vivo precision of the transfer guides used to attach the bonded spheres.
This pilot study is the first to assess accuracy intraorally using an industrial grade scanner as a reference with accuracy and precision beyond that achievable with IOS or conventional impressions. It is not limited to the subject’s dentate situation and can depict the full buccal surface, which makes it possible to visually evaluate variations in accuracy beyond the reference-bodies in the transfer method [ ]. The aim of this pilot study was to evaluate the methodology, where accuracy of IOS and conventional impressions could be analyzed. Further, to assess precision of intraoral scanners in vivo.
The null hypotheses were firstly that the reference scanner would demonstrate negligible intra-system deviations, thus displaying a high precision and allowing the scanner to be used as a reference scanner. Secondly, that the laboratory scanner would show similar negligible intra-system deviations. Thirdly, that accuracy between IOS and conventional impressions, and fourthly, that precision of IOS systems would show no statistically significant differences.
2
Materials and methods
2.1
Subject characteristics and treatment
Five subjects with referrals for implant treatment of a missing first or second premolar in the upper jaw participated in the study after informed consent and ethics approval (Dnr 2015/324; Regional Ethical Review Board, Uppsala). The inclusion criteria were no other missing teeth with residual spacing in the premolar to premolar area. Upon 12 weeks healing of implant surgery with two different diameters, (4 x RP 4.3 and 1 x NP 3.75, Nobel Parallel CC; Nobel Biocare AB, Gothenburg, Sweden), healing abutments (4 x RP and 1 x NP, both 5 mm diameter; Nobel Biocare AB), were removed and scan-bodies were connected to the implants and hand-tightened, (4 x Elos Accurate Intra Oral 2A-B Scan Body, 1 x Elos Accurate Intra Oral 2A-A Scan Body; Elos Medtech, Gothenburg, Sweden).
2.2
Scans
Subjects were positioned in an operating chair raised at 30° angle with the head and neck fixated using an orthopedic vacuum pillow (223940000; Camp Scandinavia AB, Helsingborg, Sweden). The chair was stabilized with dorsal supports to eliminate movement.
Clear self-retractors were placed (Adult Self Retracting; Photomed, Van Nuys, USA) to create a clear and dry field of view. Four reference-bodies in the shape of hemispheres in Alumina (Al 2 O 3 ) with refraction index 1.72, (Goodfellow, Huntingdon, UK), milled to 3 mm in diameter, (Kullberg Mikroteknik AB, Lycke, Sweden), were bonded with light-curing luting cement, (Variolink Esthetic LC; Ivoclar Vivadent, Schaan, Lichtenstein) and LED cured, (Bluephase Style; Ivoclar Vivadent) to primarily the buccal enamel of laterals and second premolars in the upper jaw. However, adaptations had to be made due to anatomical and physical variations, and the reference-body was shifted in three cases unilaterally. In one case the reference-body was bonded to the first molar positioned in the space of a missing second premolar. In a second case, the reference-body was bonded to the first premolar as the second premolar was the actual implant position. In the third case, the reference-body was bonded to a central incisor, as the lateral was fully restored in porcelain. The reference-bodies were numbered as position one to four from right premolar (Pos. 1), via right lateral (Pos. 2), left lateral (Pos. 3) to left premolar (Pos. 4).
The teeth were carefully air-dried and a light coating layer was administered (CEREC Optispray; Sirona, York, USA). To evade excessive coating beyond the teeth of the upper jaw, an anterior and occlusal contraster, (Anterior Contraster and Occlusal Contraster; Photomed, Van Nuys, USA) was used to physically block coating particles from unnecessarily adhering to retractors, soft-tissues, and opposing dentition.
An industrial-grade scanner, ATOS Core 80 5MP, with software and manufacturer details listed in Table 1 , was mounted on a proprietary tripod on stable concrete flooring, to eliminate any micromovements due to vibrations during scanning sequences. The scanner unit was calibrated and tested according to VDI/VDIE 2634 (VDI e.V.; Düsseldorf, Germany), displaying maximum deviations: 1 μm probing error form (sigma), −4 μm probing error (size), 2 μm sphere spacing error and − 3 μm length measurement error. The scanner was used to scan the subjects’ upper jaw, and due to the natural curvature of the jaw, each scan consisted of 5–7 successful sequences. The scan commenced in the midline and subsequently moved laterally and posteriorly by positioning the target field in the view-finder of the stereo camera, with at least fifty percent overlapping the previous sequence. Several sequences were automatically discarded due to software warnings of movement during the sequence scan. Reference stickers normally, applied outside the scanning area to assist the transformations of each sequence in the ATOS system, could not be used due to the need for coating and the risk of displacement due to poor adherence. Therefore, transformations relied on best-fit-alignment of overlapping sections. Due to anatomical limitations, it was not possible to capture data posteriorly of the second premolar in some subjects. The setup of ATOS, coated reference-bodies and retractors can be seen in Fig. 1 .
| System | Manufacturer | Software | Light Source | Color |
|---|---|---|---|---|
| ATOS Core 80 5MP | GOM, Braunschweig, Germany | Pro 8.1 | Blue Light | Monochrome |
| 3M True Definition | 3M, St. Paul, USA | 2.0.2.0 | LED | Monochrome |
| Omnicam (CEREC) | Sirona, York, USA | 4.3 | LED | Generic color |
| Trios 3 | 3Shape, Copenhagen, Denmark | 1.3.4.5 | LED | True to color |
Coating was removed using rigorous water spray and microbrushes, (Quick Stick microbrushes; Dentonova AB, Stockholm, Sweden) and retractors were detached. Subjects were asked to rinse with water, and after a fifteen-minute recess, allowing for rehydration of teeth, the scan-body was removed and re-attached. Soft retractors were placed, (Optragate; Ivoclar Vivadent) to allow for a dry operating field. The upper jaw was scanned three times with each IOS per manufacturer protocol in the following order: Trios 3 (TRIOS), CEREC Omnicam (OMNI), and 3M True definition (3M), the latter after coating with proprietary spray applicator, (3M Powder Sprayer; 3M, St. Paul, USA), ( Table 1 ). As the 3M scanner required coating, the order of scanning was maintained as removal of all coating was considered an absolute necessity between scanners. After finalizing scans with IOS, the coating from 3M was removed using water spray and teeth were cleaned with microbrushes before removing soft retractors. Minor coating particles were allowed in non-essential areas. Scan-bodies were replaced with hand-tightened impression copings for open tray (4 x RP CC and 1 x NP CC; Nobel Biocare AB) and one impression (IMPR), (Impregum Penta Soft; 3M, St Paul, USA) in a tray, (Position Tray; 3M, St. Paul, USA) with cut-out for the impression pin, was carried out on each subject.
The impressions were disinfected (MD 520; Dürr Dental AG, Bietigheim-Bissingen, Germany) according to manufacturer’s recommendation, and gently air-dried. Impressions were poured between 24 and 48 h with type IV dental stone (Fujirock EP; GC Europe, Leuven, Belgium). All impressions prior to pouring and subsequent gypsum models were stored at room-temperature (+20 to +22 °C) for 24 h. An appropriate sized scan-body was re-attached to the model analogue and hand tightened. Each model was scanned three times with D1000 laboratory scanner, (D1000; 3Shape, Copenhagen, Denmark).
2.3
Processing scan-data and analysis
All non-essential data, such as traces of retractors, soft tissues or opposing dentition, were manually removed in all ATOS sequences to only include buccal scan data of teeth and scan-bodies.
The resulting scans were polygonised in the software with settings at high detail to create an STL file.
STL files were exported from all IOS and the dental laboratory scanner. 3M files were acquired from the 3M Online Case Manager and specific proprietary dental laboratory software was used to export STL files from OMNI, (InLab 15; Sirona), and both TRIOS and D1000 dental laboratory scanner, (Dental System 2015; 3Shape).
The ATOS reference-files, IOS files and D1000 laboratory scans of gypsum models were imported into 3D inspection and metrology software, (Geomagic Control 2015; Geomagic, North Carolina, USA). IOS files were manually cropped to remove all soft-tissues. For subsequent calculation of accuracy and precision, and after best-fit alignment based on the manual selection of the four reference-bodies excluding interfering peripheral bonding material, a 3D Compare Analysis was performed. Manual surface annotations of 2 mm diameter were selected on the top cap of the reference-bodies for deviation measurements and statistical analysis. Color histograms were included to visualize the distribution of deviations of the surfaces with settings at nominal ±20 μm and critical ±100 μm.
The use of reference-bodies for best-fit alignment over full surface match was preferred as scan acquisition was optimized for the reference-bodies regarding coating as well as angle of acquisition. Furthermore, intra-subject measurement annotations from each reference-body were gathered consistently.
2.4
Precision evaluation of ATOS reference-scanner and D1000 laboratory scanner
The three ATOS reference-scans acquired from each subject were cross-compared. Similarly, the three scans with D1000 dental laboratory scanner of one gypsum model per subject, were cross-compared.
2.5
Accuracy evaluation of IOS scans and conventional impression
The first ATOS reference-scan of each subject, (ATOS S1), was used to assess accuracy within each IOS scanner group: 3M, OMNI, and TRIOS. Accuracy was evaluated for IMPR, based on the first D1000 laboratory scan, (D1000 S1).
Initial investigations of scatterplots in R [ ] showed that non-negligible correlation was present between measurements from two positions, and hence a bivariate response had to be considered. Moreover, the variability on an individual basis implied that a random-effects model was necessary. This was managed by fitting a random-effects model with a bivariate response; one regressor in the form of the factor describing the system and a random-effect variable. The model was fitted by the routine lme in the R package nlme [ ]. Multiple comparisons were carried out by routines in the R package multcompView [ ]. Residuals of the fitted model were investigated by plotting and the fitted models were considered adequate. Significance level was set at p = 0.05.
2.6
Precision evaluation of IOS scans
The three IOS-scans for 3M, OMNI, and TRIOS, were cross-compared.
Fig. 2 shows a schematic overview of the protocol, and subsequent accuracy and precision tests.
2
Materials and methods
2.1
Subject characteristics and treatment
Five subjects with referrals for implant treatment of a missing first or second premolar in the upper jaw participated in the study after informed consent and ethics approval (Dnr 2015/324; Regional Ethical Review Board, Uppsala). The inclusion criteria were no other missing teeth with residual spacing in the premolar to premolar area. Upon 12 weeks healing of implant surgery with two different diameters, (4 x RP 4.3 and 1 x NP 3.75, Nobel Parallel CC; Nobel Biocare AB, Gothenburg, Sweden), healing abutments (4 x RP and 1 x NP, both 5 mm diameter; Nobel Biocare AB), were removed and scan-bodies were connected to the implants and hand-tightened, (4 x Elos Accurate Intra Oral 2A-B Scan Body, 1 x Elos Accurate Intra Oral 2A-A Scan Body; Elos Medtech, Gothenburg, Sweden).
2.2
Scans
Subjects were positioned in an operating chair raised at 30° angle with the head and neck fixated using an orthopedic vacuum pillow (223940000; Camp Scandinavia AB, Helsingborg, Sweden). The chair was stabilized with dorsal supports to eliminate movement.
Clear self-retractors were placed (Adult Self Retracting; Photomed, Van Nuys, USA) to create a clear and dry field of view. Four reference-bodies in the shape of hemispheres in Alumina (Al 2 O 3 ) with refraction index 1.72, (Goodfellow, Huntingdon, UK), milled to 3 mm in diameter, (Kullberg Mikroteknik AB, Lycke, Sweden), were bonded with light-curing luting cement, (Variolink Esthetic LC; Ivoclar Vivadent, Schaan, Lichtenstein) and LED cured, (Bluephase Style; Ivoclar Vivadent) to primarily the buccal enamel of laterals and second premolars in the upper jaw. However, adaptations had to be made due to anatomical and physical variations, and the reference-body was shifted in three cases unilaterally. In one case the reference-body was bonded to the first molar positioned in the space of a missing second premolar. In a second case, the reference-body was bonded to the first premolar as the second premolar was the actual implant position. In the third case, the reference-body was bonded to a central incisor, as the lateral was fully restored in porcelain. The reference-bodies were numbered as position one to four from right premolar (Pos. 1), via right lateral (Pos. 2), left lateral (Pos. 3) to left premolar (Pos. 4).
The teeth were carefully air-dried and a light coating layer was administered (CEREC Optispray; Sirona, York, USA). To evade excessive coating beyond the teeth of the upper jaw, an anterior and occlusal contraster, (Anterior Contraster and Occlusal Contraster; Photomed, Van Nuys, USA) was used to physically block coating particles from unnecessarily adhering to retractors, soft-tissues, and opposing dentition.
An industrial-grade scanner, ATOS Core 80 5MP, with software and manufacturer details listed in Table 1 , was mounted on a proprietary tripod on stable concrete flooring, to eliminate any micromovements due to vibrations during scanning sequences. The scanner unit was calibrated and tested according to VDI/VDIE 2634 (VDI e.V.; Düsseldorf, Germany), displaying maximum deviations: 1 μm probing error form (sigma), −4 μm probing error (size), 2 μm sphere spacing error and − 3 μm length measurement error. The scanner was used to scan the subjects’ upper jaw, and due to the natural curvature of the jaw, each scan consisted of 5–7 successful sequences. The scan commenced in the midline and subsequently moved laterally and posteriorly by positioning the target field in the view-finder of the stereo camera, with at least fifty percent overlapping the previous sequence. Several sequences were automatically discarded due to software warnings of movement during the sequence scan. Reference stickers normally, applied outside the scanning area to assist the transformations of each sequence in the ATOS system, could not be used due to the need for coating and the risk of displacement due to poor adherence. Therefore, transformations relied on best-fit-alignment of overlapping sections. Due to anatomical limitations, it was not possible to capture data posteriorly of the second premolar in some subjects. The setup of ATOS, coated reference-bodies and retractors can be seen in Fig. 1 .
