A Comparative Study of the Accuracy of Dies Made from

Digital Intraoral Scanning vs. Elastic Impressions: An In Vitro

Study
Mahmoud Serag, BDS,1 Talal al Nassar, BDS,2 Dane Avondoglio, MS,3 & Saul Weiner, DDS3
1
Division of Prosthodontics, Department of General Dentistry, School of Dental Medicine, East Carolina University, Greenville, NC
2
Department of Prosthetic Dental Science, King Saud University College of Dentistry, Riyadh, Saudi Arabia
3
Department of Restorative Dentistry, Rutgers School of Dental Medicine, Newark, NJ

Keywords
Digital impressions; dies; fixed
prosthodontics; dimensional accuracy.
Correspondence
Saul Weiner, Department of Restorative
Dentistry, Rutgers School of Dental Medicine,
110 Bergen St., Newark, NJ 07103. E-mail:
Weiner@sdm.rutgers.edu.
Submitted to the Faculty of the Department
of Restorative Dentistry at Rutgers School of
Dental Medicine in partial fulfillment of the
requirements for the Certificate in
Prosthodontics and Master of Dental
Sciences
The authors deny any conflicts of interest
Abstract
Purpose: To compare the dimensional accuracy of dies generated from digital intrao-
ral scans with dies generated from conventional polyvinyl siloxane (PVS) impression
material.
Materials and Methods: A machined metal die was impressed 10 times using PVS,
and scanned 10 times using the iTero Cadent system and 10 times using the LAVA
C.O.S. system. Dies generated from each system were imaged in a standardized holder
using a microscope and measured with analytical software at three sites and compared
with the dimensions of the master die.
Results: The dimensional differences between the master die and both the PVS and
iTero groups were approximately 60 µm in all three dimensions but the Lava C.O.S.
die dimensions measured close to 100 µm at two of the three sites.
Conclusion: PVS/Die and iTero groups offered superior reproduction of the master
die in terms of dimensional accuracy, but these differences are likely of little clinical
significance.

Accepted December 17, 2015

doi: 10.1111/jopr.12481

Computer-based engineering is being incorporated in restora-
tive dentistry using digital processes from intraoral scanning
(i.e., image capture), through fabrication of the definitive
restoration. These processes have been termed the “digital
workflow.”1 Using developed algorithms, the digital workflow
can be entirely virtual or may include traditional technologies.
In a completely digital workflow, the abutment tooth or implant
image is recorded with a laser-based scanning device, a virtual
cast then developed, and the restoration designed virtually and
the digital image saved in an STL file. The file is then sent to a
multi-axis milling machine or a rapid prototyping (RP) device,
such as a 3D printer, to produce the restoration; however, in a
second algorithm, particularly where custom design is needed,
an intraoral scan of the abutment tooth or implant with a scan-
ning body is used to generate an image, an STL file, and a
resin cast is produced. The cast is articulated and used to fabri-
cate the restoration using currently available technician-based
laboratory procedures.2
Two widely used commercially available systems for intrao-
ral scanning are iTero (Align Technologies, San Jose, CA) and
Lava C.O.S. (3M ESPE, St Paul, MN). The iTero system uses
confocal imaging to capture a 3D digital image of the tooth
shape from a laser beam reflection. Confocal imaging results
in a sharp image with very precise edges, necessary for the
identification of the marginal areas of the virtual image of the
prepared tooth. This image is produced by filtering the beam
through a series of pinholes resulting in parallel light rays. No
reflective medium is required, and a series of snapshots are
“stitched” together. The image is reviewed for completeness,
and the finish line of the preparation outlined by a technician.
The image is forwarded to the milling center where a die is
milled from polyethylene. The software for production of the
working cast and die requires a full-arch virtual image. Addi-
tionally, a scan of both arches in occlusion is required.
The Lava COS system uses real-time video capture to image
a prepared tooth. Here the image is digitized continuously until

88 Journal of Prosthodontics 27 (2018) 88–93 

C 2016 by the American College of Prosthodontists
Serag et al Digital Intraoral Scanning vs. Elastic Impression

complete. A reflectance spray is required for the imaging se-

quence, and the image is recorded from light-beam reflectance.
Missing areas of the image can be rescanned and, based upon
adjacent landmarks, “patched” into the virtual image of the
arch. In addition, scans of the opposing occlusion and the ar-
ticulation are required to complete the virtual image. Then the
finish line of the prepared tooth is demarcated. The image is
then transferred via modem to an RP center for production of
the working cast and die from a polymer resin.
Mounting evidence indicates that the digital workflow can
produce restorations with clinically acceptable fits.3-5 However,
optimization of the digital process is complex in view of the
series of steps required to transfer information from the shape
and position of the abutment tooth to the systems associated
with manufacture of the restoration. An acceptable restoration
is dependent upon the accuracy of the scanning system, the
precision of the software that manipulates the digital image,
and the manufacturing process for the crown. While completely
digital workflows may optimize the results, the use of a partial
digital workflow is realistic because of the costs associated with
the purchase of milling systems and the still-common use of
metal ceramic restorations made with the traditional lost-wax
technique.6
A key aspect of digital workflow is production of the working
cast from a virtual image. Since the clinician is ultimately re-
sponsible for treatment, aspects of the digital workflow require
Figure 1 The die, identified by an arrow, is seated in the left half of the
analysis and comparison with current techniques. Accordingly,
impression tray with 3 mm of relief to allow a uniform layer of impression
the purpose of these series of experiments was to compare the
material.
accuracy of dies generated from virtual images captured with
commonly available scanning systems to those made from tra-
ditional elastomeric impressions.

Materials and methods

Die fabrication
A machined, metal master die with approximate dimensions of
occlusal diameter of 6 mm, gingival diameter of 8 mm, and a
height of 8 mm was made. The master die was centered in a
stainless steel mounting jig with a screw from its under surface
(Fig 1). Three groups of working dies were made from the
master die (Fig 2), and each group had ten specimens.
Group I included stone dies fabricated from poly (vinyl silox-
ane) (PVS) impressions. To create the working dies, a split
stainless steel tray was used to make an impression of the mas-
ter die.7 This reduced the possible distortion of the custom tray Figure 2 The impression mold is positioned in the disassembled steel
(Figs 1 and 3). The tray was designed with a 3 mm circum- impression tray.
ferential relief to allow a uniform thickness of the impression
material and incorporated four vent holes for extrusion of ex- directions, vibrated into the impression mold, and allowed to
cess impression material with side grooves to aid in the retention set for 1 hour. The steel tray was then disassembled and the
of the impression material. Tray adhesive (Coltene/Whaledent, die carefully retrieved from the impression material. Dies were
Cuyahoga Falls, OH) was painted onto each half of tray and stored in a dry environment to prevent any distortion.
allowed to dry for 10 minutes before the impression was made. Group II dies were made from digital impressions obtained
The tray was then assembled, and the screws holding the two with the iTero scanner. In this case, the master die was in-
parts of the tray together were tightened. Medium-body PVS serted into the socket of a molar tooth in a mandibular typodont
impression material (GC America, Alsip, IL) was used to make (Columbia Dentoform, Long Island City, NY) that otherwise
each of the ten impressions, and the impression material was contained unprepared ivorine teeth (Fig 4). A digital scan was
allowed to set for 7 minutes. All the impressions were poured obtained from the cast, and the image was carefully checked in
with type 5 dental stone (Modern Materials, Haraeus Kulzer- the monitor to ensure that all aspects of the die were present
Align, South Bend, IN), mixed according to the manufacturer’s in the digital image. The image was then sent to the central

Journal of Prosthodontics 27 (2018) 88–93 

C 2016 by the American College of Prosthodontists 89
Digital Intraoral Scanning vs. Elastic Impression
Figure 3 Examples of the dies produced from the different impression
techniques are shown. From left to right are an elastomeric impression,
iTero scan, and Lava C.O.S. scan.
Figure 4 The die has been seated in the typodont in preparation for
digital scanning.
Figure 5 The die is placed in the holder on the microscope stage used to
image the dies in a reproducible manner and imaged with a magnification
of 6.25×.
90 Journal of Prosthodontics 27 (2018) 88–93 
Serag et al
Figure 6 The dimensions of the die used in these experiments are
illustrated as described in the text.
laboratory (Cadent Inc, Carlstat, NJ) to fabricate the master
die. Ten individual scans were each made and transmitted by
modem to the milling center. The dies were then milled from
polyethylene blocks. The primary author did all the scans, su-
pervised by a prosthodontist experienced in the use of the iTero
system. All dies were stored in a dry environment at room
temperature to avoid any distortion.
Group III dies were made from digital impressions using
the Lava C.O.S. system. The typodont model with the inserted
master die was used for these scans as well. The Lava system
requires the application of a titanium oxide spray on the cast
prior to the scan. In addition, the Lava system requires an artic-
ulation, so a scan of the articulated maxillary cast was obtained
after application of the titanium dioxide spray. The images were
obtained by the primary author under the supervision of a den-
tist who is a trainer for the Lava C.O.S. system. The images
were sent by modem to 3M ESPE for fabrication of 3D–printed
stereolithographic dies from a polymer resin. The dies were
carefully stored in a dry environment at room temperature.7
Die measurements
To obtain reproducible measurements, a positioning cradle was
fabricated from plaster and placed on the microscope stage,
allowing the master die and the working dies to be oriented in
the same manner for measurements, (fig 5). Imaging was done
using a microscope (Olympus America, Central Valley, PA) at a
magnification of 6.25×, and the image captured digitally (DEL-
750; Optronics, Fremont, CA). The image was transferred to a
microcomputer, and measurements of the dies were made with
analytical software (Bioquant, Nova; BIAC, Nashville TN).
Measurements were made for each die at three dimensions:
r DI: the mesiodistal width of the dies at the occlusal sur-
faces;
r DII: the occlusogingival height of the dies;
r DIII: the mesiodistal width of the dies at the margins
(Fig 6).
For each dimension, three measurements were made and av-
eraged to obtain a dimensional mean. The mean measurement
of the ten specimens in the group at each site was averaged,
and the mean and standard deviation were calculated. Mea-
surements of the three sites were also obtained from the master
die.
C 2016 by the American College of Prosthodontists
Serag et al Digital Intraoral Scanning vs. Elastic Impression

Table 1 Master die measurement and mean value of the three die Table 2 Mean and standard deviation of the absolute differences be-
systems (in µm) tween master die and test specimenss. Same letter no statistical differ-
ence, B is larger than A
Dimension I Dimension II Dimension III
N Mean SD p Post hoc
N Mean SD N Mean SD N Mean SD
DI PVS 10 44 33 0.52 No difference
Master Die 6152 8095 8162 iTero 10 44 20
PVS 10 6160 47 10 8109 32 10 8175 38 Lava 10 55 48
iTero 10 6187 34 10 8033 54 10 8125 46 DII PVS 10 28 18 0.016 A
LAVA 10 6199 57 10 8171 93 10 8271 59 iTero 10 63 53 A
Lava 10 95 69 B
DIII PVS 10 35 17 0.017 A
iTero 10 51 29 A
To identify possible experimental errors associated with the Lava 10 108 58 B
measurement process, three measurement series were done.
The first was to identify errors associated with the parallax
phenomenon, the second with placement of the specimens on
the microscope stage, and the third with the reproducibility of Discussion
the measurement process. To determine the accuracies of the
measurement process, ten measurements were made of one of The results of this study confirm that digital scanning can be
the sites of the master die, and mean and a standard deviation used to produce clinically acceptable working dies for fixed
calculated. To determine the reproducibility of placement of prosthodontic restorations. The errors observed were not of
the die in the cradle, the master die was taken out and replaced clinical significance in most circumstances.
in the cradle ten times and the measurement site recorded ten Nevertheless, strategies that may minimize differences in the
times. The mean and standard deviations were calculated. The working die compared to the tooth can improve the predictabil-
magnitude of the standard deviation was used to evaluate the ity of the digital process. One variable that may be important
errors of the measurement process. is the spray application necessary for the Lava C.O.S. system.
To calculate the accuracy of the dies generated from the Dehurtevent et al showed that variations in the spray technique
three experimental groups at each of the measurement sites, can influence the dimensions of the imaged die.8 With regard
the dimensions at each site of the working die were subtracted to the iTero system, in a comparative analysis of dies produced
from that of the master. These differences were collated and by four techniques, Hwang et al showed that dies milled from
organized by group, and means and standard deviation obtained. polyethylene blocks had the greatest dimensional errors when
Because the data were not normally distributed, a non- compared to digital images of the die, RP dies, and stone dies.9
parametric method was also used for data analysis of the dif- The standard deviations observed in this study serve to rein-
ferences in die dimension between the master die and the three force the need to more thoroughly understand the nuances of
experimental groups. The Kruskal-Walllis test was used to com- the digital workflow. It has long been recognized with regard to
pare the measurements among the groups at each site, and a the lost-wax technique that the various steps in the process, the
Mann-Whitney U-test was used for the post hoc testing. positive-negative transformations, must be organized in such a
way that the dimensional changes associated with each level in
Results the process combine to result in the “output:” a casting that has
a marginal gap within a clinically acceptable range. For pro-
The dimensions of the master die were: D1, the diameter of duction of the working cast from a digital workflow the steps
the die at its occlusal surface: 6152 µm; D2, the height of include intraoral scanning, the production of the STL virtual
the prepared die: 8095 µm; and D3, the diameter at the base image, and the manufacture of the actual working die. Each of
of the preparation: 8162 µm. The mean values of the dies in these steps has some variation. The experimental error in this
groups 1 (impression), 2 (iTero), and 3 (Lava) are reported in study was calculated to be +20 µm. Cuperus et al observed
Table 1. The differences in dimension at the three sites between errors in the STL image of a tooth that ranged from 20 to
the master die and the dies generated by the three techniques 40 µm.10 Lei and Hsu11 and Bohez et al12 reviewed the 5-axis
were compared at each of the three sites with a non-parametric milling process and reported milling errors of up to +2.0%. In
ANOVA and post hoc pairwise comparisons. These are reported addition, in the formation of the STL image, the tessellation
in Table 2. Statistical analysis showed that there was no statis- process rounds off 2D and 3D line angles. Difficulties also ex-
tical difference between the three groups at site DI. At sites ist in milling of line angles in these 3D structures. Silva et al
DII and DIII, pairwise comparisons did not show any statisti- reported that 3D printing and selective laser sintering of max-
cal difference between the PVS and iTero systems; however, illofacial models had dimenstional errors of 2.15% to 2.6%.13
a statistical difference was demonstrated for the Lava C.O.S. The literature has provided information regarding both the
system. The Lava system group dies were significantly larger reliability and accuracy of the digital workflow process used
than the other groups, p < 0.016 at site DII and p < 0.017 at in restorative dentistry. These studies can be categorized into
site DIII; however, the clinical significance is minimal since the three groups: the first is an evaluation of the scanned digital
net effect would be to provide more space for cement. image (i.e., the impression), the second is the accuracy of the

Journal of Prosthodontics 27 (2018) 88–93 

C 2016 by the American College of Prosthodontists 91
Digital Intraoral Scanning vs. Elastic Impression
working cast, either virtual or actual, and the third is the fit of
restorations produced from the digital workflow.
For an evaluation of the digital image or computer-aided im-
pression (CAI), both Patzelt et al14 and Ender and Mehl15,16
have used reference industrial scanners whose reliability and
accuracy have been previously established. The data sets from
intraoral scanners including CEREC, Lava C.O.S., iTero, and
Zfx Intra Scan (Zimmer Dental, Carlsbad, CA) were compared
statistically and clinically to conventional elastomer-based im-
pression techniques.14,17 These studies demonstrated that while
there were statistical differences and in fact the conventional
systems demonstrated greater accuracy, the dimensions of the
die obtained from both impression techniques were well within
the range of clinical acceptability.14-17 In these studies, the
term “trueness” was substituted for accuracy and “precision”
for reliability.15 This series of studies reported trueness within
40 µm of the data recorded from the reference scanner.
With regard to the casts generated by CAI techniques, studies
have examined both virtual and real casts. Virtual casts have
been extensively studied in the orthodontic literature where
storage of pre- and post-treatment casts needed for treatment
planning and case evaluation is imperative.18-20 Storage of these
casts requires much space in a busy orthodontic practice. The
dimensional accuracy of the casts has been evaluated using
Geomagic software (Morrisville, NC). A comparative analysis
suggests that the dimensional accuracy of the casts was within
one-third of a millimeter. A comparison of RP with milling
suggests that the former results in a more accurate cast for
studies in which an actual cast was produced.20 Studies in the
prosthodontic literature have reported trueness in the range of
30 to 50 µm for actual casts. In these studies as well, casts made
by RP techniques showed improved precision on the order of
20 to 40 µm when compared to milled casts. These findings
are in agreement with our findings as well as with Luthardt et
al’s.21
The majority of studies evaluate the digital workflow by
assessing the final restoration relative to fit and marginal adap-
tation. In a study by Patzelt et al, a series of crowns was
made, the intaglio surfaces of the crowns scanned, and the
virtual images superimposed on virtual images of the dies to
assess discrepancies.22 Such discrepancies could result from
errors in the scanning process or perhaps from the manufac-
turing process. In the case of milled restorations, dullness in
the bur or vibrations associated with the milling process may
occur. As noted previously, in the case of RP, errors in the
edges of the layerings, termed tessellation or polymerization
shrinkage, during curing of the resin may be the sources of
discrepancy between the die and the restoration if the die was
scanned.
The majority of these studies, however, used measurement of
marginal discrepancy as the criterion for clinical acceptability.
Retention or stability of the crown on the die was not assessed,
although this factor is of great clinical importance. These stud-
ies have all reported pre-cementation marginal gaps less than
120 µm, a commonly accepted standard, although not clinically
validated.
While there is evidence that digital workflows can pro-
duce restorations clinically equivalent to those from classic
prosthodontic protocols, aspects of these workflows may sig-
92 Journal of Prosthodontics 27 (2018) 88–93 
Serag et al
nificantly influence the outcome and must be examined. While,
as noted above, errors have been identified, little attempt has
been made to identify systematic errors: errors that routinely
occur and for which a standardized compensation is required.
Such errors may include placement of the scanning device, the
number of photos included in the stitch, and the identification
of the finish line in the digital image.23 The use of linear, 2D
measurements may be an effective way to identify such types of
errors quantitatively, and this study has identified such errors.
As an example of such errors that have been previously iden-
tified, the use of die spacers for crowns made by the lost-wax
technique is based on systematic research of the need to pro-
vide an improved post-cementation seal and retention.24 One
study that has approached the digital work flow systematically
is dimensional accuracy of full arch casts by Patzelt et al.14
However, they did not examine the dimensions of the individ-
ual die. Instead they studied the positional relationships of the
dies in a full-arch cast. Further work is needed to examine the
digital workflow and its output.
Conclusions
Casts made from a partial digital workflow have a dimensional
accuracy equivalent to that of stone casts made from traditional
impressions. While dimensional differences are present, their
magnitude is similar to that observed with stone casts and epoxy
casts.
References
1. Ng J, Ruse D, Wyatt C: A comparison of the marginal fit of
crowns fabricated with digital and conventional methods. J
Prosthet Dent 2014;112:555-560
2. Christensen GJ: Impressions are changing: deciding on
conventional, digital or digital plus in-office milling. J Am Dent
Assoc 2009;140:1301-1304
3. Syrek A, Reich G, Ranftl D, et al: Clinical evaluation of
all-ceramic crowns fabricated from intraoral digital impressions
based on principal of active waveform sampling. J Dent
2010;38:553-559
4. Giannetopoulos S, van Noort R, Tsitrou E: Evaluation of the
marginal integrity of ceramic copings with different marginal
angles using two different CAD/CAM systems. J Dent
2010;38:980-986
5. Al-Rababah MA, McFarFarlane TV, McCord JF: Vertical
marginal and internal adaptation of all ceramic copings made by
CAD/CAM technology. Eur J Prosthodont Res Dent 2008;
16:109-115
6. Lee SJ, Gallucci GO: Digital vs conventional implant
impressions: efficiency outcomes. Clin Oral Implants Res
2013;24:111-115
7. Gordon GE, Johnson GH, Drennon DG: The effect of tray
selection on the accuracy of elastic tray materials. J Prosthet
Dent 1990;63:12-15
8. Dehurtevent M, Robberecht L, Behin P: Influence of dentist
experience with scan spray systems used in direct CAD/CAM
impressions. J Prosthet Dent 2015;113:17-21
9. Hwang YC, Park YS, Kim HK, et al: The evaluation of working
casts prepared from digital impressions. Oper Dent
2013;38:655-662
C 2016 by the American College of Prosthodontists
Serag et al
10. Cuperus AMR, Harms MC, Rangel FA, et al. Dental models
made with an intraoral scanner: a validation study. Am J Orthod
Dentofacial Orthop 2012;142:308-313
11. Lei WT, Hsu YY: Accuracy test of five-axis CNC machine tool
with 3-D probe-ball. Part I: design and modeling. Int J Machine
Tools Manufacture 2002;42:1153-1162
12. Bohez ELJ, Ariyajunya B, Sinlapeecheewa C, et al: Systematic
geometric rigid body identification of 5-axis milling machines.
Computer-Aided Design 2007;39:229-244
13. Silva DN, Gerhardt de Oliveira M, Meurer A, et al: Dimensional
error in selective laser sintering and 3D-printing of models for
craniomaxillary anatomy reconstruction. J Craniomaxillofac
Surg 2008;36:443-449
14. Patzelt SB, Emmanouilidi A, Stampf S, et al: Accuracy of
full-arch scans using intraoral scanners. Clin Oral Investig
2014;18:1687-1694
15. Ender A, Mehl A: Accuracy of complete-arch dental
impressions: a new method of measuring trueness and precision.
2013;109:121-128
16. 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
17. Seelbach P, Brueckel C, Wostmann B: Accuracy of digital and
conventional impression techniques and workflow. Clin Oral
Investig 2013;17:1759-1764
Journal of Prosthodontics 27 (2018) 88–93 
C 2016 by the American College of Prosthodontists
Digital Intraoral Scanning vs. Elastic Impression
18. Bell A, Ayoub AF, Siebert P: Assessment of the accuracy of a
three-dimensional imaging system for archiving dental study
models. J Orthod 2003;30:219-223
19. Hazeveld A, Huddlestom Slater JJ, et al: Accuracy and
reproducibility of dental replica models reconstructed by
different rapid prototyping techniques. Am J Orthodont
Dentofacial Orthop 2014;145:108-115
20. Kim JH, Kim KB, Kim WC, et al: Accuracy and precision of
polyurethane dental arch models fabricated using a
three-dimensional subtractive rapid prototyping method with an
intraoral scanning technique. Korean J Orthod 2014;44:69-
76
21. Luthardt RG, Loos R, Quaas S: Accuracy of intraoral data
acquisition in comparison to the conventional impression. Int J
Comput Dent 2005;8:283-294
22. Patzelt SB, Bishti S, Att W, et al: Accuracy of computer-aided
design/computer-aided manufacturing-generated dental casts
based on intraoral scanner data. J Am Dent Assoc
2014;145:1133-1140
23. van der Meer WJ, Andriessen FS, Wismeijer D, et al:
Application of intra-oral dental scanners in the digital workflow
of implantology. PLoS One 2012;7:e43312
24. Emtiaz S, Goldstein G: Effect of die spacers on precementation
space of complete-coverage restorations. Int J Prosthodont
1997;10:131-135
93