Statement of problem
Complete-arch digital scans are becoming popular as digital dentistry is adopted for expanded clinical situations such as complete-arch prostheses, removable prostheses, extensive implant-supported treatment, and orthodontic aligners. Whether the scan pattern technique affects the trueness and precision of complete-arch scans and whether differences in accuracy exist among different scanners remain unclear. Furthermore, each manufacturer recommends a different scan pattern, but evidence of the superiority of the manufacturer’s recommended pattern is lacking.
The purpose of this in vitro study was to determine whether the scan pattern affects the trueness, precision, and speed of complete-arch digital scans performed by using 4 different digital scanning systems.
Material and methods
A custom model used as the reference standard was fabricated with teeth having the same refractive index as dentin and enamel to simulate the natural dentition. The scan of the custom typodont was obtained by using an ATOS III Triple Scan 3D optical scanner. This study evaluated the CEREC Omnicam, Planmeca Emerald, Align iTero Element, and 3Shape TRIOS 3. Experimental scans were obtained from each of the 4 different digital scanning systems by using 4 unique scan patterns by experienced clinicians. Four experimental scans were acquired from each of the scanners by using 4 distinct scan patterns for a total of 16 scans for each scanner. Scan patterns 1 to 4 were based on the operator manuals for each different scanner. The scan time was recorded for each scan. All experimental scans were converted to standard tessellation language (STL) format, and a comprehensive metrology program, Geomagic Control X, was used to compare the reference standard scan with the experimental scans.
For trueness, the scanner ( P <.001), scan pattern ( P =.001), and their interaction ( P <.001) were found to be significant. Overall, scan pattern 2 showed the highest average trueness and precision. Likewise, for overall scan pattern precision, the scanner, scan pattern, and their interaction were found to be significant ( P <.001).
Scan pattern affected trueness and precision for some scanners, but not for others. Differences exist in the complete-arch scan speed, trueness, and precision of individual scanners. Scan pattern can play an important role in the success of digital scanning.
Scan pattern affects the trueness, precision, maximum deviations, and time of complete-arch scans.
Interest in intraoral scanners (IOSs) has increased in dentistry since the introduction of the CEREC system in 1984. The rapid pace of innovation in digital scanners has increased the indications for use to include more complicated complete-arch prostheses, removable prostheses, extensive implant-supported prostheses, and orthodontic aligners. Intraoral digital scans are now acceptable for complete arch prostheses supported by nonparallel implants, which is a testament to how far the technology has advanced. These expanded indications combined with an increase in the ease of use have increased the adoption of digital workflows. This results in increased patient comfort, shortened treatment time, and the potential for decreased overhead, all while maintaining the quality of conventional systems. Additionally, intraoral digital scans are equal to if not better than conventional techniques regarding trueness, precision, and prosthetic quality. As new scanners and software upgrades are released, an evaluation of their accuracy is essential to ensure continued progress.
Variables such as scanning technology, blood or saliva contamination, and scanning pattern can limit IOS accuracy. Accuracy as defined by International Organization for Standardization (ISO) standards consists of both trueness and precision ( Fig. 1 ). Trueness is defined as the amount a test object or data set deviates from a reference object (reference standard) or data set ( Fig. 1 ). A scanner with higher trueness delivers a 3D object rendition that more closely matches the originally scanned object. Precision represents the repeatability of measurements. A scanner with higher precision delivers more consistent results after repeated scans ( Fig. 1 ).
Scan strategy and its effects on IOS accuracy have been evaluated. In studies published in 2013, Ender and Mehl tested 3 IOS systems by using 5 scanning strategies. They reported that although the IOS devices were capable of high accuracy, close to a previously reported 20.4 μm as seen in conventional polyvinyl siloxane impressions, the scan strategy used affected the results. For smaller segments such as sextants, the scan pattern did not affect trueness or precision. Mennito et al evaluated sextant scans from 6 IOS devices. In the study, 5 distinct scan strategies were evaluated, and minimal discrepancies were noted among scanners. However, the scan pattern has been shown to influence trueness and precision when larger areas such as a complete arch are scanned. Müller et al evaluated complete-arch maxillary digital scans by using the TRIOS Pod scanner. Three scan strategies were evaluated from digital scans of a stone cast. They reported differences in accuracy among the scan patterns tested.
The purpose of this in vitro study was to evaluate the effect of 4 scan patterns on the trueness and precision of 4 different digital scanners. The study compared multiple different scanners by using a specially fabricated typodont with a refractive index closely matching that of actual teeth. The null hypothesis was that the scan pattern would not affect the trueness or precision of each scanner and that no differences would exist among scanners. The knowledge gained from this research could help clinicians choose the scan pattern yielding the most accurate results for the IOS system used.
Material and methods
Four IOS systems were assessed in this study: Planmeca Emerald (PE; Planmeca) software v5.9.4, 3Shape TRIOS 3 color model (TR; 3Shape) software v184.108.40.206, iTero Element (IE; Align Technology) software v220.127.116.111, and CEREC Omnicam (CO; Dentsply Sirona) hardware v2.24 and software v4.5.2. A custom reference standard model was fabricated by using the method reported by Mennito el al. Briefly, typodont teeth were prepared and restored by using composite resin (Telio CAD; Ivoclar Vivadent AG) (TC), which has a refractive index (RI) of 1.49. The RI of TC is similar to that of dentin (1.54) and enamel (1.63), thus simulating natural tooth substance. The reference standard model was obtained by using an optical scanner (ATOS III Triple Scan 3D; gom). The ATOS is a noncontact structured blue-light scanner that works by using multiple cameras that record the projection of stripes on an object being measured. For each pixel of the camera sensor’s points, coordinates can be estimated with high precision. For jaw-sized scans, this scanner has a trueness of 3 μm and precision of 2 μm.
Four scan strategies were chosen for this study based on manufacturer-recommended patterns for the scanners. To decrease the risk of operator bias and error, each IOS was tested by an operator who was familiar with and trained on each particular device with a minimum of 3 years of experience. Four experimental scans were obtained from each of the scanners for each of the 4 distinct scan strategies. Scan patterns (SPs) 1 to 4 were based on the IE, CO, PE, and TR operator manuals, respectively. The experimental groups for this study were as follows: SP1, SP2, SP3, and SP4 for CO with a sample size of 4 repeated scans for each pattern for a total of 16 scans. The same SP groups were repeated with EI, PE, and TR. For each scan performed, both scan time and rendering time were recorded. Scan strategies are shown in Figure 2 . One operator (J.L.) was designated as a calibrator and was present during all scanning sessions to ensure that the appropriate scan patterns were used and that the time was recorded accurately. Furthermore, a calibration session demonstrating the appropriate scan patterns was held with all clinicians before the study.
All experimental scans were converted to standard tessellation language (STL) format by the corresponding manufacturers’ recommended conversion method. A comprehensive metrology program (Geomagic Control X; 3D SYSTEMS) was used to compare the reference standard model with the experimental STL models.
Once imported into the software, the models were digitally trimmed along a reference line made on the original solid model by using the software’s trim function ( Fig. 3 ). Once models were trimmed, the reference standard model STL file was imported and trimmed on the opposite side of the original reference line to ensure adequate test model overlap ( Fig. 4 ).
Using Geomagic’s Initial Alignment and Best Fit Alignment functions, the models were overlaid in preparation for 3D comparisons ( Fig. 5 ). The software’s Best Fit Alignment function acts to align the test file and reference file by using an iterative closest point algorithm (ICP). ICP has become one of the most widely adopted methods for aligning digital 3D files. It is a data-driven approach that uses point cloud properties to aid in aligning 3D objects. The algorithm uses 3D correspondences between 2 clouds of points and determines the minimal distance between objects. The software then evaluates the test file, and the points that are the closest are computed on the reference file ( Fig. 6 ).
The software’s 3D compare function allowed for customization of measurements and color mapping of results. A value of 0.5 mm was used to define the upper and lower limits for color mapping ( Fig. 6 ). No reference tolerances were set for this study, and color mapping was chosen to display deviations on the digital models. Models displayed after the 3D compare function showed a range of colors that correlated with potential areas of mismatch between the test and reference models. Areas highlighted in darker blue indicated a negative or inward deviation, whereas areas highlighted in darker red indicated a positive or outward deviation of the test models. Reports were generated for each comparison, and the average, minimum deviation, maximum deviation, and standard deviation values were compiled in a spreadsheet (Excel; Microsoft Corp) ( Fig. 6 ).
Descriptive statistics are presented as means and standard deviations. For each outcome, generalized linear mixed models with scanner, scan pattern, and their interaction in the model were used for trueness, precision, maximum deviation, and scan time. A random intercept was included in the model to account for replicates. Normality assumptions were checked to ensure no transformations were needed. Post hoc pairwise comparisons were made when appropriate by using a Scheffé adjustment (α=.05, except when the Scheffé adjustment was used). All statistical analyses were performed by using a statistical software program (SAS v9.4; SAS Institute Inc).
Scanners were first evaluated overall for trueness, precision, and time averages. This included all 16 scans performed for each IOS device. The scanners were also compared in 2 main ways: scan patterns within each scanner ( Table 1 ) and scanners within each scan pattern ( Table 2 ).
|Pattern||Scanner||N||Trueness (Average Micrometer)||Trueness Rank (Within Scanner)||Precision (STD Micrometer)||Precision Rank (Within Scanner)||Scan Time (Minutes)||Scan Time Rank (Within Scanner)|