The purpose of this study was to assess the reproducibility of in-vivo and ex-vivo scans using an intraoral scanner.
Twenty adults with no missing teeth except for third molars were included in the study. Alginate impressions were taken, and plaster models were made from the impressions. Each subject underwent full-arch intraoral scanning twice with a TRIOS scanner (3Shape, Copenhagen, Denmark) at an interval of 2 weeks, and, the plaster models were scanned at the same interval with the same scanner. The first images of each scan were superimposed on the second scanned images using surface-based registration. In each case, the differences between the 2 scanned images were evaluated with color mapping. The reproducibility between the in-vivo and ex-vivo scans was compared using independent t tests and Bland-Altman analysis.
The discrepancies between the first and second images were greater in the posterior than in the anterior regions for both the in-vivo and ex-vivo scans. Average surface differences between the first and second images were greater for the in-vivo scans (0.04 mm) than for the ex-vivo scans (0.02 mm). The Bland-Altman plots showed that the reproducibility of both scans was within the limits of agreement.
The reproducibility of in-vivo scanning was comparable with ex-vivo scanning, although it showed a slight difference (0.02 mm) compared with ex-vivo scanning.
In-vivo intraoral scanning showed a 0.041-mm average mean difference between 2 scans.
Ex-vivo intraoral scanning showed a 0.019-mm average mean difference between 2 scans.
Reproducibility of in-vivo scanning was similar to that of ex-vivo scanning.
With the advances in computer technology, digital dental models are now being widely used for orthodontic diagnosis and treatment planning. The use of digital models alleviates many obstacles and challenges of plaster models made from conventional impressions, including the burden of storage, the risk of damage or breakage, and the difficulties in sharing the data with other clinicians involved in the patients’care. Digital dental models can be created through either indirect or direct techniques. Indirect methods involve laser scanning or computed tomographic imaging of the alginate impressions or plaster models, and direct methods involve intraoral scanners. Recently, with the introduction of chair-side intraoral scanners, interest in obtaining digital dental models using the direct method has increased.
Several researchers have reported on the reproducibility of intraoral scanners. Wiranto et al and Naidu and Freer showed that tooth-width measurements on digital models and intraoral scans did not differ significantly from the caliper measurements taken from the corresponding plaster models. Akyalcin et al reported that tooth-width measurements from intraoral scans were in near-perfect agreement with the caliper measurements on dry skulls and were more accurate than the cone-beam computed tomographic measurements. Hayashi et al found that intraoral scanners can be used to generate accurate and reliable digital dental models by comparing it with other desktop scanners. Similarly, Grünheid et al reported that the digital models made using the chair-side oral scanner did not differ significantly from the plaster models. They suggested that despite undesirable chair time and patient acceptance, chair-side oral scanning is clinically acceptable. Su and Sun also reported that the precision of scanning with an intraoral scanner was clinically acceptable, although the level of precision decreased as the scanning scope increased.
To investigate the accuracy of intraoral scanners, previous studies used caliper measurements on plaster models or on dry skulls as the gold standard, or scans of dental models made from conventional impressions. However, there are limitations to the use of plaster models as the gold standard: eg, distortion during model fabrication and inaccuracies related to the impression materials. In addition, the authors of all the studies that included scanned plaster models as the reference did not use the same scanner for intraoral scanning and for scanning the plaster models.
Recently, Flügge et al evaluated the repeatability of 10 intraoral scans using an iTero scanner and compared this with the reliability of 10 extraoral scans of a plaster model from 1 subject. Although they used an identical scanner for both in-vivo (intraoral) and ex-vivo (extraoral) scans, only 1 subject, which might be an insufficient sample size, was used in their study. The purpose of this study was to evaluate the reproducibility of an intraoral scanner by comparing the reproducibility of in-vivo scans for 20 participants with that of ex-vivo scans of the plaster models of the same participants with the same intraoral scanner.
Material and methods
This study was approved by the institutional review board of the Chonnam National University Dental Hospital in Gwangju, Korea. All subjects who were enrolled in the study gave their informed consents to participate. The inclusion criteria were (1) a full permanent dentition from the second molar to the contralateral second molar, and (2) no metal or gold crown restorations. Twenty subjects without severe crowding and dentofacial deformity were included in this study.
All subjects had intraoral scanning with a TRIOS scanner (3Shape, Copenhagen, Denmark), according to the manufacturer’s recommendations. Before the scanning, the calibration and preheating for the scanner tip were accomplished to the scanner. The teeth were dried lightly using compressed air. Scanning was started with the mandibular dentition according to the program’s function, with the left second molar and continued to right second molar along the occlusion. First, the occlusal surfaces were scanned and then the lingual and buccal surfaces. In the maxillary arch, the occlusal surfaces were also scanned first the same as in the mandible, and then the buccal and lingual surfaces were scanned. When scanning the occlusal surfaces, the scanner head was kept at 0 to 5 mm from the teeth. For scanning the lingual and buccal surfaces, the scanner tip was rolled 45° to 90° to the lingual and buccal sides, respectively. The image could be continuously viewed on the screen during the scanning process; this allowed direct visual feedback to ensure that no areas were missed. After the arch scans, a bite scan in centric occlusion was recorded, with the buccal surfaces of both molars and premolars were included.
Alginate impressions (Cavex Impressional, Cavex Holland, Haarlem, The Netherlands) were taken from all subjects and immediately poured with dental stone (New Plastone II White, GC Corporation, Tokyo, Japan). The same examiner (J.-S.L.) scanned the plaster models with the TRIOS scanner. Both intraoral and plaster model scans were repeated after 2 weeks. All data were sent to the OrthoAnalyzer (3Shape) software program for reprocessing as STL files.
To evaluate the reproducibility of the in-vivo and ex-vivo scans, the first and second scanned images were superimposed by means of the software’s best-fit algorithm. A reverse engineering software program (Rapidform; 3D Systems, Rock Hill, SC) was used to register the 2 scanned images. The registration process was performed automatically by the software program using the “register” function. First, initial registration was done. The initial registration involved the selection of 3 corresponding points on each of the 2 images, after which the program’s automatic fine-registration function was used to finalize the matches. The 3 corresponding points for initial registration were the mesiobuccal cusp tips of the right and left second molars and the mesiolabioincisal point angle of the right central incisor. The software calculated an initial fit by 3 corresponding points between the 2 images automatically to obtain the reference and data close and register features. The initial registration established a rough initial alignment from which to start the fine registration. The software used the iterative closest point algorithm for the fine registration. Since the adjacent soft tissues could increase the range of error, these areas were deleted, along with the gingival margin, to allow superimposition of the clinical crowns.
Using the shell/shell deviation function of the program, the average surface differences between the first and second scanned images were computed at all points on the surfaces. In addition, the differences between the 2 images were evaluated by means of the color-mapping methods ( Fig 1 ).