Accuracy of Intraoral Scans An in Vivo Study of Di
Accuracy of Intraoral Scans An in Vivo Study of Di
Accuracy of Intraoral Scans An in Vivo Study of Di
Supported by a grant (ORF42001) of the Oral Reconstruction Foundation, Basel, Switzerland. The funders had no role in study design, data collection, and analysis, decision
to publish, or preparation of the paper.
a
Clinician Scientist, Department of Oral and Maxillofacial Surgery, Translational Implantology, Medical Center e University of Freiburg, Faculty of Medicine, University of
Freiburg, Freiburg, Germany.
b
Anthropologist, Division of Biological Anthropology, Faculty of Medicine, University of Freiburg, Freiburg, Germany; and Post-doctorate Scientist, Department of Oral and
Maxillofacial Surgery, Translational Implantology, Medical Center e University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany.
c
Doctorate Candidate, Department of Oral and Maxillofacial Surgery, Translational Implantology, Medical Center e University of Freiburg, Faculty of Medicine, University of
Freiburg, Freiburg, Germany.
d
Master Dental Technician, Mehrhof Implant Technologies GmbH, Berlin, Germany.
e
Statistician, Institute of Medical Biometry and Statistics, Medical Center e University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany.
f
Associate Professor, Department of Prosthetic Dentistry, Medical Center e University of Freiburg, Faculty of Medicine, University of Freiburg, Freiburg, Germany.
g
Full Professor, Department of Oral and Maxillofacial Surgery, Translational Implantology, Medical Center e University of Freiburg, Faculty of Medicine, University of Freiburg,
Freiburg, Germany.
h
Full Professor, Charité e Universitätsmedizin Berlin, Corporate Member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Department of Oral and Maxillofacial
Surgery, Berlin, Germany; and Senior Research Scientist, Department of Oral and Maxillofacial Surgery, Translational Implantology, Medical Center e University of Freiburg,
Faculty of Medicine, University of Freiburg, Freiburg, Germany.
Table 1. Study population with selected jaw, regions, and number of test
surfaces
Clinical Implications
Participant Jaw Regions of Test Surfaces
The accuracy of intraoral scans is a prerequisite for 1 Mandible First premolar left, second premolars, first molars
the fabrication of dental restorations in CAD-CAM 2 Maxilla Second premolar left, first molars
dentistry. However, the longer the scanned distance 3 Mandible Second premolars
within the dental arch, the higher the inaccuracy. 4 Maxilla First premolars, second premolars, first molars
Clinical intraoral scanning did not result in clinically 5 Mandible First premolar right, second premolars, first molars
acceptable accuracy for complete-arch scans. 6 Maxilla Second premolar left, first molars
7 Mandible First premolar left, second premolar left, first molars
8 Maxilla First premolar left, second premolars, first premolars
9 Mandible First premolars, second premolars
The precision of intraoral scans in vivo has been 10 Maxilla First premolars right, second premolars, first molars
assessed by repeated scanning and comparison of the 11 Mandible First molars
resulting virtual casts.5-8 The lower precision of intraoral
scans compared with that of extraoral scans has been
reported, and intraoral conditions including moisture,
denture teeth were used to assess spatial deviations. The
salivary flow, and restricted space can complicate accu-
authors are unaware of a previous study that examined
rate intraoral image acquisition.5,7
the trueness and precision of multiple IOSs in vivo and
Trueness may be assessed in vitro by scanning an
assessed the spatial location of deviations. The research
experimental model with the respective intraoral scanner
hypotheses of the study were that the accuracy of IOSs
and a reference scanner and comparing the resulting
would be lower when used in the oral cavity and that
virtual casts. This approach has been used to assess
short-span scans would be acquired with higher preci-
trueness in vitro, but the trueness of IOSs has been re-
sion and trueness than complete-arch scans.
ported to decrease when acquiring a complete dental
arch compared with a short span.9-15
MATERIAL AND METHODS
Trueness in vivo has been approximated by
comparing an intraoral scan with a virtual reference The study included intraoral scanning of voluntary par-
model produced from a conventional impression. How- ticipants using different IOSs and an individually fabri-
ever, the conventional impression itself introduces inac- cated experimental appliance. The study protocol was
curacy,16 and other studies have used different approved by the Institutional Review Board of the
approaches to study trueness. Nedelcu et al17 used an Medical CenterdUniversity of Freiburg, Freiburg, Ger-
industrial scanner and IOSs to acquire the facial surface many (434/17), registered at the German Clinical Trial
of the maxillary anterior teeth. This method captured Register (DRKS: 00014039), and performed in accordance
only a short span and did not evaluate the molar area with the institutional research committee and the 1964
most prone to deviations or the complex acquisition of Helsinki declaration and its later amendments. Study
multiple surfaces. Atieh et al18 focused on comparing participants were recruited in the Department of Oral
conventional impressions with an intraoral scanner by and Maxillofacial Surgery, Translational Implantology in
using a reference appliance made from metal alloy. The the University Medical Center Freiburg during July 2017
deviation of intraoral scans was measured only in small and July 2018. Participants gave written consent to the
areas distributed throughout the dental arch. Notably, in study before inclusion.
previous studies, the scans were aligned by using a best- Participants with partial edentulism in both posterior
fit algorithm including the complete surface before the regions of either the maxilla (n=6) or the mandible (n=5)
assessment of the deviations. With this method, the were selected (Table 1). This specific anatomic situation
deviations are arbitrarily averaged over the entire surface was required for the placement of the experimental
and their spatial location is not reflected. This becomes appliance and the arrangement of scan objects. Table 1
especially important when scanning larger areas, as po- displays the selected jaw and number and regions of
tential inaccuracies at the end of a long span are averaged the test surfaces. Participants with limited mouth open-
over the entire arch. This can be overcome by using a ing or hard and soft tissue defects, including extended
different method of alignment as implemented in the scar tissues in the oral cavity, were excluded.
current study. The experimental appliance was manufactured by
An experimental appliance comprising denture teeth using a pink acrylic resin base plate, denture teeth, and
and reference bodies was fabricated and scanned intra- cylindrical reference bodies. A stone cast of the respective
orally with multiple scanners, as well as extraorally with a jaw of each participant was poured from a conventional
reference scanner. Alignment of multiple scans was irreversible hydrocolloid (Pluralgin Super; Pluradent)
selectively performed by using reference bodies, and the impression. A visible light-polymerized denture base
Figure 1. A, Individually fabricated experimental appliance with reference objects (R1, R2, R3) and test surfaces represented by acrylic resin denture
teeth (T). B, Occlusal view of inserted appliance.
material (Megatray Basisplatte; Megadenta Dentalpro- saliva ejector were applied. Passivity was confirmed and
dukte GmbH) was adapted to cover the dentate and the appliance was fixed by using a denture adhesive
edentulous areas of the arch. A geometric pattern of in- (Blend-a-dent Plus Haftcreme; Oral-B) Intraoral scans
dentations was created on the denture base material to were acquired by the same operator using the 3 scanning
achieve a morphology that was easily captured by the devices, and 3 scans were made with each IOS. Caution
surface scanner. Cast denture teeth (SR Ivocron; Ivoclar was used to scan only the intraoral device and not the
Vivadent AG) were fixed with an adhesive (Palapress; participant’s teeth because of the risk of movement of the
Kulzer GmbH) to the base plate in both the premolar and experimental appliance. The scheme for data acquisition
molar areas. Three cylindrical reference bodies (Strau- and evaluation are displayed in Figure 2.
mann Mono Scanbody, Art.032.041; Institut Straumann Standard tessellation language (STL) files of the refer-
AG) were fixed with the adhesive (Palapress) in the area ence scans, extraoral scans, and intraoral scans were im-
of the second molars bilaterally and in the anterior region ported into a 3D modeling software program (Meshmixer;
of the baseplate (Fig. 1). The experimental appliance was Autodesk Inc). The reference bodies and test surfaces were
designed to imitate the intraoral anatomy, including the cropped to prepare the scan for evaluation. The reference
test surface and the pink acrylic resin base plate, and host bodies were used for the registration of multiple scans and
reference bodies for data registration and evaluation. were therefore essential for the assessment of the spatial
A reference data set of the experimental appliance location of the deviations. Registration was defined as the
was established with a desktop optical scanner (S600 3D alignment of multiple scans using common surface
Arti; Zirkonzahn) with a manufacturer specified precision information. The protocol for registration is displayed in
of 10 mm. The accuracy of the acquired data was eval- Table 2. To elaborate the critical mass of information for an
uated by using a coordinate measuring machine (CMM) accurate registration, a randomly chosen scan was regis-
(DS 10; Renishaw). The positions of the reference objects tered multiple times. Randomization was achieved by us-
(R1-R3) were acquired with this high precision tactile ing a simple randomization sheet in Microsoft Excel
scanner (accuracy <5 mm) and used for verification.19 (Microsoft Corp) and applying it to all members of the
Comparison of the acquired optical scanner data with study population. Registration was performed with 1
the CMM data showed a median deviation of 22 mm reference body (R1), 2 reference bodies (R1, R2), and 3
when using a best-fit registration algorithm. reference bodies (R1, R2, R3), respectively. The comparison
The experimental appliance was scanned on the stone showed no differences in registration accuracy. Therefore,
cast with the following IOSs: TRIOS3, version 1.6.10.1 the final assessment of deviations was performed by using
(3Shape A/S) (TR); CEREC Omnicam, version 4.6, 1 reference body (R1 or R3) at each distal site. The test
(Dentsply Sirona) (OC); True Definition, version 5.4 (3M) surfaces were used for the measurement of deviations
(TD). Each scan was performed by 1 operator (F.K.) using between scans.
the manufacturer’s recommended scanning path. Each The registration of only 1 selected scan body was
scanner was used 3 times for extraoral scanning. performed by using a best-fit registration algorithm
The respective experimental appliance was placed in based on an iterative closest point search procedure.20
the mouth of the participant and cheek retractors and a The scan body surfaces were first aligned according to
T2
test group was performed by using a mixed linear
Precision
Trueness
intraoral
T2 intraoral regression model and Bonferroni correction after pairwise
scan
R2
scan R1 R3
comparisons (Stata Statistical Software: StataCorp. 2017.
T1 T2 3 intraoral scans Release 15; StataCorp LLC) (a=.05).
T2
R1 R3
R2
Trueness
Reference scan
extraoral T1 T2 Precision RESULTS
scan T2 extraoral
scan
R1 R3 A total of 11 participants were enrolled in the study.
3 extraoral scans Three extraoral and 3 intraoral scans with each of the 3
IOSs resulted in 9 extraoral scans and 9 intraoral scans
Figure 2. Number of scans and comparisons for assessment of precision
and trueness.
per participant.
The median precision of extraoral scanning of short
spans was 29 mm (TD), 22 mm (TR), and 23 mm (OC)
(Fig. 3, Table 3). The median precision of extraoral
Table 2. Protocol for registration of multiple scans and evaluation of
scanning of long spans was 165 mm (TD), 81 mm (TR),
deviations between surfaces and interpretation of results with regard to
scan length
and 103 mm (OC) (Fig. 3, Table 3).
Registration Evaluation Result
The median precision for intraoral scanning of short
R1 T1 Short span
spans was 31 mm (TD), 23 mm (TR), 43 mm (OC) (Fig. 4,
R1 T2 Long span R2 Table 3). The median precision of intraoral scanning of
R3 T2 Short span T1 T2 long spans was 153 mm (TD), 80 mm (TR), and 198 mm
R3 T1 Long span T2 (OC) (Fig. 4, Table 3).
The median trueness of extraoral scanning of short
R1 R3
spans was 40 mm for TD, 28 mm for TR, and 36 mm when
using OC (Table 3). The median trueness of extraoral
scanning of long spans was 581 mm (95th percentile: 1387
mm) (TD), 132 mm (TR), and 118 mm (OC) (Table 3).
their principle axis, and then an iterative closest point The median trueness of intraoral scanning of short
search was performed. This was done by finding the spans was 47 mm (TD), 38 mm (TR), and 45 mm (OC)
closest point on the target surface. Correspondences (Fig. 5, Table 3). The median trueness of intraoral scan-
pointing in the wrong direction (normal vectors) were ning of long spans was 433 mm (TD), 147 mm (TR), and
discarded. Of the remaining corresponding points, those 198 mm (OC) (Fig. 6, Table 3).
further away than the 90th percentile of distances were The precision of the IOSs was significantly different
discarded to avoid invalid registration results. Based on for extraoral and intraoral long-span scans. Pairwise
these correspondences, the reference was iteratively comparison showed significantly higher precision for
rotated to the target surface. Finally, the transformation extraoral long-span scans with TR compared with TD
was applied to the entire scan. (P=.005) and intraoral long-span scans with TR
To evaluate the mesh discrepancies, the (unsigned) compared with OC (P<.001) and TR compared with TD
distance for each vertex in the region of interest on the (P<.001). The intraoral and extraoral precision of short-
reference mesh to the surface of the aligned scans was span scans was significantly higher than for long-span
recorded. This resulted in tens of thousands of error scans for all scanners (P<.001).
values per alignment that were averaged per vertex over The trueness of the IOSs was significantly different
all iterations and then accumulated over all participants.7 except for intraoral short-span scans (P=.87). Pairwise
Mesh alignment and error assessment were performed comparison showed lower trueness of TD compared with
by using the mathematical and statistical platform R and TR (P<.001) and OC (P<.001) for extraoral and intraoral
specifically the R-packages Rvcg, Morpho, and long-span scans, and extraoral short-span scans. The
mesheR.21-23 intraoral and extraoral trueness of short-span scans was
The primary end point was the precision of intraoral significantly higher than for long-span scans for all
and extraoral scanning with 3 different scanners, evalu- scanners (P<.001).
ated by comparing the 3 consecutive scans in each group
with 1 randomly selected scan as the reference. The
DISCUSSION
secondary end point was the trueness of intraoral and
extraoral scanning with 3 different IOSs, assessed by The results of the present clinical study supported the
aligning the 3 consecutive scans in each group with the hypotheses that the accuracy of IOS would be lower
virtual reference model. when used in the oral cavity and that short-span scans
Table 3. Median of precision and trueness of IOSs TD, OC, and TR for
extraoral and intraoral scanning
400 Precision (mm) Trueness (mm)
Short Extraoral (95th Intraoral (95th Extraoral (95th Intraoral (95th
Span Percentile) Percentile) Percentile) Percentile)
µ)
Precision (uµ
Trueness (uµ)
interquartile range.
100
600 50
Precision (uµ)
0
400 True definition TRIOS Omnicam
Figure 5. Trueness of intraoral scanning of TD, TR, and OC for short span.
Values of deviation given in mm. Boxplots do not show outliers, that is,
200 whiskers restricted to maximum length of 1.5 times interquartile range.