Three-Dimensional Static Articulation Accuracy of Virtual

Models – Part I: System Trueness and Precision

Sophia Hui Xin Yee, BDS, MDS,1 Roxanna Jean Esguerra, DDS, MS,2 Amelia Anya Qin’An Chew, BDS,
MDS,3 Keng Mun Wong, BDS, MSD,4 & Keson Beng Choon Tan, BDS, MSD 4

1
Department of Dental Surgery, Khoo Teck Puat Hospital, Singapore, Republic of Singapore
2
Private Practice, Denver, CO
3
Department of Restorative Dentistry, National Dental Centre of Singapore, Singapore, Republic of Singapore
4
Faculty of Dentistry, National University of Singapore, Singapore, Republic of Singapore

Keywords
Coordinate measuring machine; model
scanner systems; precision; static articulation
accuracy; trueness.
Correspondence
Keson B. Tan, Faculty of Dentistry, National
University of Singapore, 11 Lower Kent Ridge
Road, Singapore 119083, Republic of
Singapore. E-mail: dentanbc@nus.edu.sg
Keson Beng Choon Tan ORCID
http://orcid.org/0000-0002-5530-6838.
The authors deny any conflicts of interest
related to this study.
Accepted September 16, 2017
doi: 10.1111/jopr.12723
Abstract
Purpose: To evaluate the 3D static articulation accuracy of 3 model scanner-CAD
systems (Ceramill Map400 [AG], inEos X5 [SIR], Scanner S600 Arti [ZKN]) using a
coordinate measuring machine (CMM). Trueness and precision for each system will
be reported in Part I.
Materials and Methods: The master model simulated a single crown opposing a
3-unit fixed dental prosthesis. Five mounted stone cast sets were prepared, and one set
was randomly selected. Reference values were obtained by measuring interarch and
interocclusal reference features with the CMM. The stone cast set was scanned 5 times
consecutively and articulated virtually with each system (3 test groups, n = 5). STL
files of the virtual models were measured with CMM software. dRR , dRC , and dRL ,
represented interarch global distortions at right, central, and left sides, respectively,
while dRM , dXM , dYM , and dZM represented interocclusal global and linear distortions
between preparations.
Results: For trueness values, mean interarch global distortions ranged from 13.1 to
40.3 μm for dRR , –199.0 to –48.1 μm for dRC , and –114.1 to –47.7 μm for dRL .
Mean percentage error of interarch distortion did not exceed 0.6%. Mean interocclusal
distortions ranged from 16.0 to 117.0 μm for dRM , -33.1 to 101.3 μm for dXM , 32.9
to 49.9 μm for dYM and –32.0 to 133.1 μm for dZM. ANOVA of trueness found
statistically significant differences for dRC , dRL , dRM , dXM , and dZM . For precision
values, absolute mean difference between the 10 superimposition combinations ranged
from 25.3 to 91.0 μm for dRR , 21.5 to 85.5 μm for dRC , 24.8 to 70.0 μm for dRL .
Absolute mean difference ranged from 49.9 to 66.1 μm for dRM , 20.7 to 92.1 μm for
dXM , 86.8 to 96.0 μm for dYM , and 36.5 to 100.0 μm for dZM . ANOVA of precision
of all test groups found statistically significant differences for dRR , dRC , dRL , dXM
and dZM , and the SIR group was the least precise.
Conclusion: The overall interarch global distortion of all three model scanner-CAD
systems was low and did not exceed 0.6%. Variations in scanner technology, virtual
articulation algorithm, and use of physical articulators contributed to the differences
in distortion observed among all three groups.

An accurate transfer of a patient’s maxillomandibular relation-
ship is important for fabrication of individualized prostheses.
This can be achieved through conventional methods or more
recently with computer-aided design and computer-aided man-
ufacturing (CAD/CAM) methods and fully digital workflows.1
The CAD/CAM workflow includes a part of the conventional
process where stone casts are first mounted on an articula-
tor, scanned, virtually articulated, and the prosthesis designed
digitally.1 Many companies have developed CAD/CAM sys-
tems with virtual articulators for this purpose.1 However, the
accuracy of the scanning process and algorithm for articulating
maxillary and mandibular virtual models have not been previ-
ously investigated. Errors incorporated in the virtual articula-
tion process of the CAD/CAM workflow can have a significant
impact on the final prosthesis and the time required for chair-
side adjustment. Tamim et al2 evaluated metal-ceramic pros-
theses fabricated from intraoral digital impressions made with
the iTero intraoral scanner and found that 80% of the crowns

Journal of Prosthodontics 00 (2017) 1–8 

C 2017 by the American College of Prosthodontists 1
Static Articulation Accuracy System Trueness & Precision Yee et al

required no occlusal adjustment. While Kollmuss et al3 found A

that all-ceramic prostheses produced by the CEREC system and C
UC
laboratory CAD/CAM system required more time for occlusal
adaptation than those fabricated with conventional technique.
Z X
Previous studies have used various 3D analysis software to
Y Z Y
measure the accuracy of model scanners. Vandeweghe et al4 O
UR UL O X
investigated the accuracy of four scanner systems (Imetric, Lava UM
ST, Smart Optics, KaVo Everest) by comparing scans of stone
casts using a 3D metrology software. Lava ST performed the
best, with a mean deviation of 47 μm, while Smart Optics LR LL
performed the worst, with a mean deviation of 207 μm. Flügge LM MP
et al5 compared the precision of the iTero intraoral scanner with
the 3Shape D250 model scanner using 3D analysis software. SA
Repeated scanning of the stone casts with the model scanner
produced the highest precision with a mean deviation of 10 μm, LC

while repeated scanning with the intraoral scanner resulted in B

a mean deviation of 25 μm. Similarly, Su and Sun6 found that
the mean deviation for the extraoral scanner group ranged from Figure 1 Master models (left, A and B) reference spheres and cones. X,
8.7 to 24.3 μm, which was significantly lower than the intraoral Y, and Z orientation of local coordinate axes shown with origin (O) defined
scanner group. The precision of the intraoral scanner decreased at the centroid of the upper reference plane. Orientation of acrylic master
when the scope of scanning increased. models (right, C) mounted on custom-made sliding apparatus (SA) with
The accuracy of a model scanner is determined by two fac- mounting plates (MP) relative to local coordinate system.
tors: trueness and precision.7 Trueness is defined as the close-
ness of agreement of the scanned object with the actual ref- A
erence, while precision refers to the closeness of agreement UC
C
between repeated scans.7
To date, there is a lack of studies investigating the 3D accu- C
racy of virtual articulation of partially dentate stone casts by Z X
different model scanner-CAD systems. The objective of this O Y
UL O Z
study was to evaluate the 3D static articulation accuracy of UR UM
X Y
three model scanner-CAD systems using a coordinate measur-
AR
ing machine (CMM). Part I aims to evaluate the trueness and
precision of the three model scanner-CAD systems. Part II will
compare the 3D static articulation accuracy of these three model LR LM
LL
scanner-CAD systems using two articulation methods.

Materials and methods

Master model
A fully dentate maxillary master model with a wide neck (WN)
solid abutment analog (6.5 mm diameter, 5.5 mm height, Strau-
mann, Basel, Switzerland) at the maxillary right second molar
position and a partially dentate mandibular master model with
a regular neck (RN) solid abutment analog (4.8 mm diameter,
4.0 mm height, Straumann) at the mandibular right second pre-
molar position and WN solid abutment analog (Straumann) at
the mandibular right second molar position simulated a clinical
scenario of a single maxillary right second molar crown and
a three-unit fixed dental prosthesis (FDP) with the mandibu-
lar right second premolar and second molar as abutments. The
master models were processed with heat-polymerized, tooth-
colored poly(methylmethacrylate) (PMMA) (Dentalon Plus;
Heraeus Kulzer GmbH, Hanau, Germany) for the areas with
dentition and pink PMMA (Lucitone 199 Denture Base Resin;
Dentsply, York, PA) for the alveolus and base. Grade 5 sili-
cone nitride 8.0 mm spheres (Cerbec; CoorsTek, Golden, CO)
with a sphericity of 0.13 μm were embedded at the upper right
(UR), central (UC), and left (UL) sites of the maxillary master
LC
B
Figure 2 Stone replicas (left, A, B) with labelled reference spheres and
cones. X, Y, and Z orientation of local coordinate axes shown with origin
(O) defined at the centroid of the upper reference plane. Orientation of
stone replicas (right, C) mounted with acrylic resin interocclusal record
(AR) on Artex CR articulator relative to local coordinate system.
model and lower right (LR), central (LC), and left (LL) sites
of the mandibular master model to act as reference points for
measuring interarch distortion. The solid abutment analogs at
the upper right second molar (UM) and the lower right second
molar (LM) were used as reference points for measuring in-
terocclusal distortion (Figs 1A and B). The acronyms for the
master models and test groups are shown in Table 1.
The master models were secured to separate mounting plates
that were machined to fit a custom-made sliding apparatus
(Fig 1C). The apparatus allowed for accurate repositioning of
the master models during the study. Occlusal equilibration en-
sured that even simultaneous occlusal contacts were present at
the remaining posterior teeth and canines. Occlusal contacts

2 Journal of Prosthodontics 00 (2017) 1–8 

C 2017 by the American College of Prosthodontists
Yee et al Static Articulation Accuracy System Trueness & Precision

Table 1 Acronyms for test groups and reference features quently stored at room temperature for 72 hours to mitigate the
effects of dental stone expansion8 (Figs 2A and B).
Reference features

Arch level Upper right UR Occlusal Upper right UM Conventional mounting

sphere level second molar
Upper central UC Lower right LM
Rigid interocclusal records were made using autopolymerizing
sphere second molar acrylic resin (GC Pattern Resin LS; GC Corp., Tokyo, Japan)
Upper left sphere UL between the FDP abutments with the opposing posterior teeth to
Lower left sphere LL articulate the stone casts for conventional mounting procedures.
Lower central LC The interocclusal record was trimmed to retain only the cusp
sphere tips and flat planes of the implant abutments. The stone casts and
Lower right LR interocclusal records were randomly assigned, articulated, and
sphere mounted on a calibrated arcon articulator (Artex CR; Amann
Model scanner-CAD system
Girrbach, Pforzheim, Germany) with Type 3 dental stone (Elite
Arti Fast; Zhermack, Rovigo, Italy) (Fig 2C). The mounted
Ceramill Map 400 (Amann Girrbach, Koblach, Austria) AG stone casts were stored at room temperature for 96 hours to
inEos X5 (Sirona, Bensheim, Germany) SIR allow for the complete expansion of the mounting stone.9 The
Scanner Arti S600 (Zirkonzahn, Gais, Italy) ZKN occlusal contacts of all the stone casts were verified with 8 μm
shimstock film and remounted if found to deviate. Five sets of
mounted stone casts were prepared for the entire study.
were checked with 8-μm shimstock film (Arti-Fol; Bausch,
Nashua, NH). The intra-arch dimensions of the master models Digital scanning
were monitored for 30 days to ensure that the models were
dimensionally stable before proceeding. The three model scanners-CAD systems used in this study were:
Ceramill Map400 (Amann Girrbach), inEos X5 (Sirona Dental
Systems), and Scanner S600 Arti (Zirkonzahn). The units tested
Conventional impressions
were normal production structured light scanners supplied and
Perforated acrylic resin custom trays (Tray Resin II; Shofu calibrated by the respective agents. All scans were performed
Inc., Kyoto, Japan) extending to the landing area were fabri- by one user according to the manufacturer’s instructions. One
cated. They were then coated with adhesive (Polyether Ad- set of mounted stone casts was randomly selected as the test
hesive; 3M ESPE, St. Paul, MN) and allowed to dry for sample for Part I and scanned consecutively 5 times by each
15 minutes. Five sets of impressions were made using polyether model scanner (Fig 3). Precautions were taken to ensure that
impression material (Impregum Penta Soft; 3M ESPE) dis- the reference features and stone surfaces of the selected set of
pensed with an auto-mixing unit (PentaMix; 3M ESPE). The stone casts remained undamaged throughout the study.
impression material was allowed to set for 6 minutes accord- Ceramill Map400 Scanner (AG) is a structured light 3D scan-
ing to manufacturer’s instructions. Type 4 gypsum stone (Silky ner that produces high resolution scan data using stripe light
Rock; Whip Mix Corporation, Louisville, KY) was vacuum projection technique. The individual stone casts were secured
mixed according to manufacturer’s instructions and poured into onto magnetic plates and scanned separately to obtain the max-
the impressions. The stone casts were separated 12 hours later illary and mandibular virtual models. The mounted models were
and checked for defects and voids. The stone casts were subse- subsequently mounted onto a calibrated system-specific fixator

Figure 3 Workflow of experimental setup for Parts I and II.

Journal of Prosthodontics 00 (2017) 1–8 

C 2017 by the American College of Prosthodontists 3
Static Articulation Accuracy System Trueness & Precision Yee et al

Figure 5 Twelve probe hits spread across three levels were taken on the
reference solid abutment analog on the master model (left) and virtual
model (right).
Figure 4 Eight probe hits were taken for reference sphere on master
model (left) and virtual model (right).

(Ceramill Fixator; Amann Girrbach), and the vestibular scan

was completed. The virtual models were virtually articulated
by the Ceramill Map400 software (V.3.2.6.691) and exported
as a single standard triangulation language (STL) file.
The inEos X5 scanner (SIR) is another 3D model scanner
that uses a five-axis robotic arm and digital stripe light pro-
jection with blue light to produce precise scanned data. The
individual stone casts were first secured onto the robotic arm
and scanned individually as the scanner captured and stitched
multiple images together to create the virtual model. A buccal
scan was completed by articulating the stone casts with the
buccal surface of the dentate side perpendicular to the scanner
camera. The inLab SW15.0 software (Sirona Dental Systems) Figure 6 Four probe hits were taken on the topmost flat plane of the
would mesh and automatically articulate the virtual models reference solid abutment analog on master model (left) and virtual model
with the buccal scan. The virtual models were exported from (right).
inLab SW15.0 software and merged into a single STL file using
an advanced 3D mesh processing software system (MeshLab
ogy), which constructed geometric spheres and cones using the
V1.3.3).
measured datum points. The local coordinate system was set up
Scanner S600 Arti scanner (ZKN) is a fully automated opti-
using UR, UC, and UL stone sphere replicas on the maxillary
cal stripe light scanner with a large measuring field that allows
stone cast (Fig 2). A low force 2.0 mm × 20.0 mm ruby probe
for the complete capture of an entire stone cast in a single scan-
(Renishaw PLC, Wotton-under-Edge, UK) took eight probe hits
ning process. The stone casts were first mounted onto the model
spread across three levels for each sphere to construct a centroid
support and scanned individually to obtain the maxillary and
(Fig 4). Centroids of UR, UC, and UL were used to construct
mandibular virtual models. The stone casts were subsequently
the upper reference plane. A line was constructed with the UR
replaced onto the Artex articulator, secured to a compatible sup-
and UL centroids and the local coordinate system defined using
port system, and mounted onto a magnetic plate in the scanner.
the plane-line-point alignment procedure with the centroid of
The buccal surfaces of the mounted casts were scanned. The
the reference plane as the origin of the local coordinate system
user identified three matching points on the individual virtual
(Fig 2). A 1.0 mm × 20.0 mm ruby probe (Renishaw PLC)
models and the buccal scan to complete the virtual articula-
was used to take three circles of four probe hits each around
tion process. The models were articulated automatically by the
UM and LM to construct a cone (Fig 5). The topmost flat plane
Zirkonzahn.Modellier program and exported as a single STL
of UM and LM was constructed by taking four hits (Fig 6).
file.
The centroid of UM and LM was the pierce point of the cen-
tral axis of the cone with the plane. The corresponding virtual
Coordinate measuring machine
constructed geometric features of reference spheres and cones
All measurements were made using a CMM (Brown and Sharpe were defined with the same virtual probe hit sequences on the
GLOBAL Silver Performance 7.10.7 CMM; Hexagon Metrol- virtual model spheres and abutment analogs.
ogy, Huntersville, NC) with a manufacturer reported accuracy The CMM software tolerance functions were used to verify
of 1.9 μm. These CMM measurements were then computed in the accuracy of each constructed geometric feature. For each
3D inspection software (PC-DMIS CAD++; Hexagon Metrol- measured feature of the stone casts, the respective flatness and

4 Journal of Prosthodontics 00 (2017) 1–8 

C 2017 by the American College of Prosthodontists
Yee et al Static Articulation Accuracy System Trueness & Precision

Table 2 Mean interarch level distortions, μm (±SD), of model scanner- 0.400

CAD systems (n = 5)
0.200
Groups dRR dRC dRL

Percentage Error (%)

0.000
AG 25.3 (22.2) −48.1 (46.2) −47.7 (25.0)
SIR 40.3 (74.2) −117.4 (72.2) −114.1 (57.5)
-0.200
ZKN 13.1 (20.8) −199.0 (17.3) −101.2 (20.3)

-0.400

roundness tolerance criterion was set at 6 μm, while the 3D

deviation of the centroids was set at 6 μm. For each measured -0.600

feature of the virtual models, the respective flatness and round-

-0.800
ness tolerance criteria were set at 15 μm, while the 3D deviation
of the centroids was set at 10 μm. The higher tolerance limit
for virtual models was chosen to compensate for minor surface Figure 7 Mean interarch level percentage error distortion, %, of model
irregularities present in the STL files. Features that exceeded scanner-CAD systems (n = 5). dRR black; dRC grey; dRL white. Error
the set tolerance criteria were rejected, and a new measurement bars indicate standard deviation.
set was made. Each feature in the stone and virtual model was
measured three times and averaged. Table 3 Mean interocclusal level distortions, μm (±SD), of model
scanner-CAD systems (n = 5)
Computation of distortion values
Groups dRM dXM dYM dZM
Distortion values were computed by calculating the difference
AG 16.0 (40.4) 101.3 (24.9) 47.0 (69.2) −32.0 (41.8)
between coordinate values of the virtual models with the ref-
SIR 39.6 (54.5) 5.8 (77.2) 32.9 (76.9) 35.0 (80.9)
erence stone model set. Interarch global distortions between ZKN 117.0 (40.8) −33.1 (16.7) 49.9 (69.3) 133.1 (29.3)
the paired right (dRR ), central (dRC ), and left (dRL ) reference
spheres were calculated. Interocclusal global distortions be-
Table 4 Absolute mean difference, μm (±SD), between superimposi-
tween UM and LM (dRM ) and linear distortions in the X-axis
tion combinations of interarch measurements of model scanner-CAD
(dXM ), Y-axis (dYM ), and Z-axis (dZM ) were calculated. For
systems (n = 10)
each test group, trueness was assessed by the mean interarch
and interocclusal distortion (n = 5), while precision was as- Groups dRR dRC dRL
sessed by superimposing and calculating the absolute mean
differences between virtual model set pairings for each test AG 27.0 (17.0) 55.5 (36.3) 30.7 (18.4)
group (n = 10 superimposition combinations). SIR 91.0 (55.1) 85.5 (58.8) 70.0 (43.8)
ZKN 25.3 (15.6) 21.5 (12.4) 24.8 (15.1)
Statistical analyses
The data were initially submitted to the Shapiro-Wilk test to distortions dXM and dZM (p = 0.002, p = 0.002, respectively).
check for the assumption of normality, and it was not rejected. ZKN had the greatest mean dRC , dRM , and dZM , which were
One-way ANOVA for independent variable SYSTEM with de- significantly different from the other two groups. AG had the
pendent variables: global distortions (dRR , dRC , dRL , dRM ) and least mean dRL and greatest mean dXM , which was significantly
linear distortions (dXM , dYM , dZM ) was performed. Student- different from the other groups.
Newman-Keuls post-hoc analysis was performed at signifi-
cance level of α = 0.05. All analyses were performed using Precision
SPSS v21.0 (IBM Corp, Armonk, NY). Absolute mean difference between the 10 superimposition com-
binations ranged from 25.3 to 91.0 μm for dRR , 21.5 to 85.5 μm
Results for dRC , and 24.8 to 70.0 μm for dRL (Table 4, Fig 8). Absolute
mean differences ranged from 49.9 to 66.1 μm for dRM , 20.7
Trueness
to 92.1 μm for dXM , 86.8 to 96.0 μm for dYM , and 36.5 to
For the three systems studied, the mean dRR ranged from 13.1 100.0 μm for dZM (Table 5, Fig 8). The overall precision of all
to 40.3 μm, mean dRC ranged from –199.0 to –48.1 μm, and three groups can be seen in Figure 8.
mean dRL ranged from –114.1 to –47.7 μm (Table 2). Figure 7 One-way ANOVA for precision between test groups revealed
illustrates these interarch distortions as percentage error of dis- significant differences for global interarch distortions dRR ,
tortion. Mean dRM ranged from 16.0 to 117.0 μm, mean dXM dRC , and dRL (p < 0.0001, p = 0.006, p = 0.003, respec-
ranged from –33.1 to 101.3 μm, mean dYM ranged from 32.9 tively) and linear distortions dXM and dZM (p < 0.0001, p =
to 49.9 μm, and mean dZM ranged from –32.0 to 133.1 μm 0.004, respectively). SIR had the greatest absolute mean dif-
(Table 3). ference for dRR , dRL , dXM , and dZM , which was significantly
One-way ANOVA for trueness between test groups revealed different from the other groups. ZKN had the least absolute
significant differences for global distortions dRC , dRL , and dRM mean difference for dRC , which was significantly different from
(p = 0.002, p = 0.006, p = 0.011, respectively) and linear SIR.

Journal of Prosthodontics 00 (2017) 1–8 

C 2017 by the American College of Prosthodontists 5
Static Articulation Accuracy System Trueness & Precision
1660
1440
Absolute Mean Difference (µm)
1220
1000
880
660
440
220
0
AG SIR
dRR dRC dRL dRM
Figure 8 Absolute mean difference, μm, between superimposition
combinations of interarch and interocclusal measurements of model
scanner-CAD systems (n = 10). Error bars indicate standard deviation.
Table 5 Absolute mean difference, μm (±SD), between superimposi-
tion combinations of interocclusal measurements of model scanner-CAD
systems (n = 10)
Groups dRM dXM
AG 50.7 (27.8) 30.6 (18.2)
SIR 66.1 (41.7) 92.1 (61.7)
ZKN 49.9 (30.4) 20.7 (12.1)
Discussion
The purpose of this study was to assess the 3D static articulation
accuracy of three model scanners-CAD systems. This study
found significant differences between the three systems at both
interarch and interocclusal levels.
Most studies to date have used various 3D analyzing soft-
ware to superimpose the images of the virtual models and
calculate the deviations. Wriedt et al,10 investigating the pre-
cision of aligning virtual casts with the vestibular scanning
technique using a photo-optical scanner, Activity 102 (Smart
Optics, Bochum, Germany), found that the greatest deviations
between the virtually articulated models averaged 37 ± 28 μm.
Solaberrieta et al11 validated the articulation of the mandibular
cast to the maxillary cast on a virtual articulator using three
reverse engineering software programs (GOM Inspect, Rapid-
form, Geomagic) and found that the virtual articulation process
resulted in a mean deviation of 69 μm.
In contrast with these previous studies, the use of a CMM
allows accurate original coordinate measurements in 3D space
to define the linear and rotational displacements that can occur
in the X, Y, and Z axes.12 The use of a partially dentate model
simulating a common clinical situation with tooth preparations,
compared to previous studies on fully dentate models, could
explain the greater amount of distortion seen. Nonetheless, the
interarch global distortion of all three systems remained low
and did not exceed 0.6%, validating the use of model scan-
ners as part of the CAD/CAM workflow for similar clinical
scenarios.
All three model scanner-CAD systems employed different
methods of articulating virtual models. For the AG scanner, the
6 Journal of Prosthodontics 00 (2017) 1–8 
Yee et al
ZKN
dXM dYM dZM
dYM dZM
86.8 (47.6) 52.3 (28.9)
94.5 (56.9) 100.0 (58.7)
96.0 (54.1) 36.5 (20.7)
Figure 9 Representative images of scans by Ceramill R
Map400 scanner
(A, B), inEos X5 scanner (C, D) and S600 Arti scanner (E, F).
maxillary and mandibular virtual models were articulated auto-
matically by the Ceramill R
Mind software, which meshed the
individual arch scans with the entire buccal vestibular scan. For
the SIR scanner, the virtual models were articulated automati-
cally by the inLab SW 15.0 software, which meshed the buccal
surfaces of the posterior teeth on one side of the virtual mod-
els with images captured during the buccal registration stage.
The ZKN scanner was the only system that did not mesh the
virtual models automatically but required the user to manu-
ally select three matching points between the virtual models
with the entire buccal scan before the virtual models were ar-
ticulated by the Zirkonzahn.Modellier software. Solaberrieta
et al13 evaluated virtual occlusal record section dimensions and
their influence on spatial positioning accuracy using an indus-
trial scanner (ATOS Compact Scan 5M; GOM, Mils, Austria).
They found that the meshing of frontal and bilateral sections re-
sulted in greater articulation accuracy. This supports the present
study, which found that the use of a unilateral buccal scan with
the SIR scanner resulted in poorer precision when compared
with the other scanners, which used a full-arch buccal scan.
In addition, variations in scanner technology and the meshing
algorithm of the associated software could have contributed
to the significant differences observed among the three model
scanner-CAD systems.
C 2017 by the American College of Prosthodontists
Yee et al
Ender and Mehl14 assessed the precision of intraoral scan-
ners by superimposing test samples within the same group and
calculating the absolute mean difference between these super-
impositions. A group with 5 test samples would yield 10 su-
perimposition combinations. Apart from SIR, the precision of
the other two groups were comparable to the mean deviation
of 37 ± 28 μm seen Wriedt et al’s study.10 The poorer level of
precision seen in SIR could be attributed to variations in scan-
ning and meshing algorithms. During the buccal registration
stage with the inEos X5 scanner, the articulator was stabilized
manually by the operator with the buccal surfaces perpendic-
ular to the scanner while the scanner captured multiple buccal
scan images that were stitched together by the inLab SW15.0
software. Unsatisfactory scan images as a result of poor po-
sitioning or movement of the articulated casts were deleted
and the process repeated until acceptable buccal scan images
were obtained. The use of an articulator plate that allowed the
articulator to remain stationary could have minimized variabil-
ity during the buccal registration stage. In comparison with
the other two scanner systems, the SIR scanner was the only
one with an open operating table that could have exposed the
scanner and stone cast to variations in environmental factors
such as ambient lighting and temperature. The use of MeshLab
software was found to have an insignificant effect on preci-
sion, as an earlier feasibility study determined deviations of
–5 to 8 μm, which was within the set tolerance criterion of
10 μm.
Beside errors incorporated during the scanning and virtual
articulation process, the use of physical articulators could have
contributed to the distortion observed. Tan et al15 investigated
the 3D repositioning accuracy of single articulator mounting
plates onto the lower member of articulators, and found that
black mounting plates used with the Artex CR articulator had
a repositioning accuracy of 8.4 to 8.9 μm in the anteroposte-
rior direction, 5.0 to 8.9 μm in the mediolateral direction, and
1.6 to 3.5 μm in the vertical direction. In contrast, a feasibil-
ity study conducted previously found that the repositioning of
both mounted casts on the articulator and closure of articulator
resulted in a greater interarch distortion, of 9 to 25 μm, which
could have affected the 3D articulation accuracy for all three
scanner-CAD systems.
This study was limited by the model scanner-CAD systems
available for testing and the logistics of scan repetitions obtain-
able. Further studies should investigate other model scanner-
CAD systems or the scenario of fabricating longer span pros-
theses.
A qualitative evaluation of the STL files obtained from the
scanning of stone casts revealed that the scans produced by
the AG scanner and ZKN scanner had a generalized stippled
appearance, while scans produced by the SIR scanner appeared
the smoothest (Fig 9). Nonetheless, the scans by all three sys-
tems were deemed to be acceptable with minimal defects and
blebs. All three systems were easy to operate with a manageable
learning curve and intuitive workflow. The SIR scanner was the
most versatile with an open operating table that allowed for the
placement of various articulators with different dimensions. In
contrast, the AG scanner was compatible only with Amann
Girrbach Artex articulators. Despite this limitation, the manu-
facturers claimed that the use of an Artex articulator with the
Journal of Prosthodontics 00 (2017) 1–8 
C 2017 by the American College of Prosthodontists
Static Articulation Accuracy System Trueness & Precision
Ceramill Fixator allowed for a 1:1 transfer of the mounted casts
to the scanner. This transfer formed part of the link in the Digital
Functional Prosthetics workflow,16 which allowed information
to be transferred in a precise manner. The ZKN scanner had an
intermediate level of versatility, as different articulators could
be mounted in the scanner by using a compatible articulator
support.
Conclusions
Part 1 of this study investigated the 3D static articulation accu-
racy of three model scanner-CAD systems and found significant
differences in trueness and precision between all three systems
at the interarch and interocclusal levels. Within the limitations
of this study, the following conclusions were made:
1. The overall interarch global distortion of all three model
scanner-CAD systems was low and did not exceed 0.6%.
2. The SIR group was significantly less precise than the AG
and ZKN groups at the interarch and interocclusal levels.
3. Variations in accuracy of scanner, virtual articulation al-
gorithm, and use of physical articulators contributed to
the differences in distortion observed among all three
groups.
Acknowledgments
The authors gratefully acknowledge the support of Straumann,
Amman Girrbach, Sirona, and Zirkonzahn for supporting this
research with their components and technical expertise, and Mr.
Chan Swee Heng, Senior Laboratory Officer, National Univer-
sity of Singapore, for his technical assistance.
References
1. Solaberrieta E, Minguez R, Etxaniz O, et al: Improving the
digital workflow: direct transfer from patient to virtual
articulator. Int J Comput Dent 2013;16:285-292
2. Tamim H, Skjerven H, Ekfeldt A, et al: Clinical evaluation of
CAD/CAM metal-ceramic posterior crowns fabricated from
intraoral digital impressions. Int J Prosthodont 2014;27:331-337
3. Kollmuss M, Kist S, Goeke JE, et al: Comparison of chairside
and laboratory CAD/CAM to conventional produced all-ceramic
crowns regarding morphology, occlusion, and aesthetics. Clin
Oral Investig 2016;20:791-797
4. Vandeweghe S, Vervack V, Vanhove C, et al: Accuracy of optical
dental digitizers: an in vitro study. Int J Periodontics Restorative
Dent 2015;35:115-121
5. Flugge TV, Schlager S, Nelson K, et al: Precision of intraoral
digital dental impressions with iTero and extraoral digitization
with the iTero and a model scanner. Am J Orthod Dentofacial
Orthop 2013;144:471-478
6. Su TS, Sun J: Comparison of repeatability between intraoral
digital scanner and extraoral digital scanner: an in-vitro study. J
Prosthodont Res 2015;59:236-242
7. British Standards Institution. 12836: 2012 EI. Dentistry —
Digitizing Devices for CAD/CAM Systems for Indirect Dental
Restorations—Test Methods for Assessing Accuracy. UK, BSI
Standards Publication, 2012
8. Heshmati RH, Nagy WW, Wirth CG, et al: Delayed linear
expansion of improved dental stone. J Prosthet Dent
2002;88:26-31
7
Static Articulation Accuracy System Trueness & Precision
9. Michalakis KX, Stratos A, Hirayama H, et al: Delayed
setting and hygroscopic linear expansion of three gypsum
products used for cast articulation. J Prosthet Dent 2009;102:313-
318
10. Wriedt S, Schmidtmann I, Niemann M, et al: Digital 3D image of
bimaxillary casts connected by a vestibular scan. J Orofac Orthop
2013;74:309-318
11. Solaberrieta E, Otegi JR, Goicoechea N, et al: Comparison of a
conventional and virtual occlusal record. J Prosthet Dent
2015;114:92-97
12. Nicholls JI: The measurement of distortion: theoretical
considerations. J Prosthet Dent 1977;37:578-586
8 Journal of Prosthodontics 00 (2017) 1–8 
Yee et al
13. Solaberrieta E, Arias A, Brizuela A, et al: Determining the
requirements, section quantity, and dimension of the virtual
occlusal record. J Prosthet Dent 2016;115:52-56
14. Ender A, Mehl A: In-vitro evaluation of the accuracy of
conventional and digital methods of obtaining full-arch dental
impressions. Quintessence Int 2015;46:9-17
15. Tan MY, Ung JY, Low AH, et al: Three-dimensional
repositioning accuracy of semiadjustable articulator cast
mounting systems. J Prosthet Dent 2014;112:932-941
16. Amann Girrbach: DFP Brochure-Digital Functional Prosthetics.
http://pdf.medicalexpo.com/pdf/amann-girrbach/dfp-brochure-
digital-functional-prosthetics/71298-130225.html
C 2017 by the American College of Prosthodontists