You are on page 1of 8


Precision of intraoral digital dental impressions

with iTero and extraoral digitization with the
iTero and a model scanner
Tabea V. Flu€ gge,a Stefan Schlager,b Katja Nelson,c Susanne Nahles,d and Marc C. Metzgere
Freiburg and Berlin, Germany

Introduction: Digital impression devices are used alternatively to conventional impression techniques and
materials. The aims of this study were to evaluate the precision of digital intraoral scanning under clinical
conditions (iTero; Align Technologies, San Jose, Calif) and to compare it with the precision of extraoral
digitization. Methods: One patient received 10 full-arch intraoral scans with the iTero and conventional
impressions with a polyether impression material (Impregum Penta; 3M ESPE, Seefeld, Germany). Stone
cast models manufactured from the impressions were digitized 10 times with an extraoral scanner
(D250; 3Shape, Copenhagen, Denmark) and 10 times with the iTero. Virtual models provided by each
method were roughly aligned, and the model edges were trimmed with cutting planes to create common
borders (Rapidform XOR; Inus Technologies, Seoul, Korea). A second model alignment was then performed
along the closest distances of the surfaces (Artec Studio software; Artec Group, Luxembourg, Luxembourg).
To assess precision, deviations between corresponding models were compared. Repeated intraoral scanning
was evaluated in group 1, repeated extraoral model scanning with the iTero was assessed in group 2,
and repeated model scanning with the D250 was assessed in group 3. Deviations between models were
measured and expressed as maximums, means, medians, and root mean square errors for
quantitative analysis. Color-coded displays of the deviations allowed qualitative visualization of the
deviations. Results: The greatest deviations and therefore the lowest precision were in group 1, with mean
deviations of 50 mm, median deviations of 37 mm, and root mean square errors of 73 mm. Group 2 showed a
higher precision, with mean deviations of 25 mm, median deviations of 18 mm, and root mean square errors of
51 mm. Scanning with the D250 had the highest precision, with mean deviations of 10 mm, median deviations
of 5 mm, and root mean square errors of 20 mm. Intraoral and extraoral scanning with the iTero resulted in de-
viations at the facial surfaces of the anterior teeth and the buccal molar surfaces. Conclusions: Scanning
with the iTero is less accurate than scanning with the D250. Intraoral scanning with the iTero is less accurate
than model scanning with the iTero, suggesting that the intraoral conditions (saliva, limited spacing) contribute
to the inaccuracy of a scan. For treatment planning and manufacturing of tooth-supported appliances, virtual
models created with the iTero can be used. An extended scanning protocol could improve the scanning
results in some regions. (Am J Orthod Dentofacial Orthop 2013;144:471-8)

Resident, Division of Oral and Maxillofacial Surgery, University Medical Center,
Freiburg, Germany. or the introduction of computer-aided design and
Research fellow, Biological Anthropology, University Medical Center, Freiburg, computer-aided manufacturing technologies in
Germany. dentistry, virtual models of teeth are required.
Professor, Division of Oral and Maxillofacial Surgery, University Medical Center,
Freiburg, Germany. Digital processes are applied for prosthetic-driven
Associate professor, Division of Oral and Maxillofacial Surgery, Charite Campus backward planning of implant surgery,1,2 orthodontic
Virchow, Berlin, Germany.
measurements, and treatment planning3-6 combined
Associate professor, Division of Oral and Maxillofacial Surgery, University Med-
ical Center, Freiburg, Germany. with surgical planning.7 Data acquired by intraoral
All authors have completed and submitted the ICMJE Form for Disclosure of scanning, computed tomography, cone-beam computed
Potential Conflicts of Interest, and none were reported. tomography, and extraoral surface scanning can be
Reprint requests to: Tabea V. Fl€ ugge, Division of Oral and Maxillofacial Surgery,
University Medical Center Freiburg, Hugstetter Str 55, 79106 Freiburg, Germany; fused.1,2,7
e-mail, For the acquisition of digital images of teeth,
Submitted, November 2012; revised and accepted, April 2013. different procedures have been described: digitization
Copyright Ó 2013 by the American Association of Orthodontists. of plaster casts,3,5,8-12 digitization of impressions,8,13 and intraoral digital impressions.13,14 The accuracy
472 Fl€
ugge et al

Fig 1. Synopsis of model manufacturing and virtual model generation.

of the different image acquisition methods and were made with 2 laser scanners (group 2, iTero; group
systems has been examined with extraoral 3, D250; Fig 1).
reference models.9,12,15-18 However, to date, no studies According to the study protocol, the data acquisition
concerning the practical application and precision of for group 1 was based on 10 intraoral scans with the
digital scanning in vivo have been done. iTero of the maxilla and the mandible of 1 patient
Digital work flow has been proposed to improve treat- (n 5 20).
ment planning, give higher efficiency, and allow new For groups 2 and 3, a unique model was used. The
methods of production and new treatment concepts.19-21 model fabrication for the extraoral scans was performed
Data storage and reproducibility are facilitated,9,19 and as follows.
treatment documentation and communication between Using the same patient as in group 1, polyether im-
professionals as well as between dentists and patients pressions of the maxilla and the mandible were taken us-
have become more convenient.22 ing a monophase polyether material (Impregum Penta;
Currently, there are a few major digital impression 3M ESPE) and stainless steel impression trays
devices: iTero (Align Technologies, San Jose, Calif), (M1W-Rim-Lock; M1W Dental, B€ udingen, Germany).
Lava COS (3M ESPE, Seefeld, Germany), and Trios The impressions were disinfected and poured with type
(3Shape, Copenhagen, Denmark) for image acquisition; IV stone (picodent U 180; Picodent, Wipperf€ urth,
and CEREC AC (Sirona, Bensheim, Germany) and E4D Germany) after a setting time of 4 hours. The first
(D4D Technologies, Richardson, Tex) for digital imaging impression was used for the production of the stone
and in-office manufacturing.14 Excluding the iTero and casts, independently of the subjective assessment of
the Trios, all scanning devices need drying and the quality. One stone cast of the maxilla and 1 stone
powdering of intraoral surfaces (CEREC, E4D, Lava cast of the mandible were made.
COS). This limits their practicability and accuracy The stone casts were scanned with the iTero using the
because powder application can add to the measuring same scanning protocol as for the intraoral scans. The
error.23 With all devices mentioned, digital impressions scans of each stone cast model (maxilla and mandible)
are acquired without contact to the gingival tissues. were repeated 10 times and produced the data set for
The precision of intraoral and extraoral scanning group 2 (n 5 20).
with the iTero as well as extraoral scanning with a model The virtual models for group 3 (n 5 20) were
scanner was examined in this study. collected by repeated scanning (10 times) of the stone
casts with a model scanner (D250).
All scans with the iTero were recorded by the same
MATERIAL AND METHODS examiner (T.V.F.) in a predetermined order. Scanning
Impressions were acquired according to the study started with the most distal tooth in the third quadrant
protocol that was approved by the ethics committee of continuing to the anterior teeth (Figs 2 and 3). Next,
the medical faculty of Freiburg University, after we the fourth quadrant was scanned, again beginning
received written consent from the study participant. with the most distal tooth. Scanning of the maxilla
One participant with a Class I occlusion and full denti- started with the most distal tooth in the first quadrant
tion was examined. and ended at the central incisor. The second quadrant
In-vivo (intraoral) scans (group 1) and ex-vivo was recorded starting with the most distal tooth. Each
(extraoral) scans of 1 patient and the patient’s models tooth was scanned from its buccal and lingual sides by

September 2013  Vol 144  Issue 3 American Journal of Orthodontics and Dentofacial Orthopedics
ugge et al 473

Fig 2. Intraoral scanning and the iTero-rendered stereolithographic model of the scanned jaw.

Fig 3. Extraoral scanning of a stone cast and the iTero-rendered stereolithographic model of the cast.

placing the camera at an angle of 45 to the tooth axis. for drawing the cutting planes. A first cutting plane
Images of each tooth showed neighboring parts of adja- running through the deepest point of the gingival sulcus
cent teeth. These served to overlap the pictures to create of the canines and the second molars in the maxilla and
a model of the whole arch from single images. All models the canines and the first molars in the mandible was
were exported in a Standard Tesselation Language (STL; created. A second cutting plane was created running
3D Systems, Rock Hill, SC) format and were used for through the transverse fissure of both second molars.
evaluation, independently of the subjective assessment. All surfaces were cut with the common cutting planes
For digitization, the stone casts were placed into to create equal basal and posterior borders (Fig 5).
the D250 scanner next to a laser source and 2 high- For quantitative analysis of precision, deviations be-
resolution cameras. During the scanning process, the tween the vertices of the surfaces were measured. Operated
platform moves the model; therefore, the laser reaches scan data were imported into Artec Studio software
the model from multiple angles. Light planes are (version; Artec Group, Luxembourg, Luxembourg)
projected onto the model, and the cameras capture their to perform a pairwise rigid body registration. Correspond-
reflections from the surface (Fig 4). ing models for each comparison were roughly aligned
The principle of triangulation was used for the manually and then registered onto another using the
creation of a 3-dimensional model, available as a stereo- implemented surface mapping algorithm.
lithographic data set. Deviations between aligned models were analyzed
All stereolithographic data sets of 1 dental arch and 1 using the software package Morpho, version 0.25 (based
scanning method (n 5 10) were imported in a common on R; created by Stefan Schlager, Freiburg, Germany).24
coordinate system and aligned by a procedure with the To estimate the differences between the surfaces, each
closest distance between 2 surfaces (Rapidform XOR; vertex on the test surface was projected to the closest
Inus Technologies, Seoul, South Korea). The models point on the corresponding control surface, and the
were orientated toward the occlusal plane to fit a view Euclidean distance was recorded. The model rendered

American Journal of Orthodontics and Dentofacial Orthopedics September 2013  Vol 144  Issue 3
474 Fl€
ugge et al

Fig 4. Scanning of a plaster cast with the D250 and rendered stereolithographic model of the cast.

Fig 5. Stereolithographic data set obtained from scan-

ning with reference planes (reference planes 1 to 3) for
cutting, displayed with Rapidform XOR.

from the first scan served as the control surface for the
consecutively acquired models in each group.

Statistical analysis
Further statistical analysis was performed with
the software R.25 For testing differences between
the groups’ distributions of averaged distances, the
Kolmogorov-Smirnov test—a distribution free and Fig 6. Colored presentation of the deviations between
nonparametric procedure—was applied. The level of surfaces in group 1.
significance was set at 0.05.
For the assessment of error, maximum, mean, and on average 57 mm (median, 43 mm); the mandible
median deviations were calculated for each group based deviated on average 43 mm (median, 31 mm). Maximum
on the averaged errors of each observation. The deviations were on average 1.137 mm in the maxilla and
distances between the vertices of the corresponding 717 mm in the mandible. Deviations of the mandible
models were displayed with color maps, so that areas were significantly lower than deviations in the maxilla.
of high and low agreement could be identified. Deviations between models are displayed in Figure 6.
The highest deviations were observed at the palatal
RESULTS borders, the facial surfaces of the anterior teeth, and the
In group 1, the virtual models, rendered from serial molars on both sides of the maxilla. In the mandible, the
intraoral scans of the maxilla and the mandible with highest deviations were at the buccal side of the molars
the iTero, were compared. The mean deviation was and the facial side of the anterior teeth. There were also
50 mm (median, 37 mm). Deviations in the maxilla were deviations above average in the interdental spaces.

September 2013  Vol 144  Issue 3 American Journal of Orthodontics and Dentofacial Orthopedics
ugge et al 475

Fig 7. Colored presentation of the deviations between Fig 8. Colored presentation of the deviations between
surfaces in group 2. surfaces in group 3.

Fig 9. Colored scale for the deviations shown in Figures 6, 7, and 8 (mm).

In group 2, the virtual models acquired with the 10 mm (median, 6 mm). Average deviations were 11 mm
repeated scans of the stone casts with the iTero were in the maxilla and 9 mm in the mandible. The maximum
compared. Deviations from the test surface were on deviation was 460 mm for the maxilla and the mandible.
average 25 mm (median, 18 mm). The maxillae had a The color-coded values of the deviations are shown in
mean deviation of 30 mm (median, 18 mm), whereas the Figure 8.
mandibles had a mean deviation of 21 mm (median, 17 Deviations in group 3 were lower than deviations in
mm). The color-coded deviations are depicted in Figure 7. groups 1 and 2. The areas of deviations showed similar
The highest deviations in the maxilla were found at patterns in the maxilla and the mandible. The inter-
the facial surfaces of the anterior teeth in the left maxilla dental spaces had the greatest deviations between the
and at the palatal borders of the models. The maximum virtual models. Deviations in the mandible were signif-
deviation was 1.79 mm in the maxilla. The mandible had icantly lower than those in the maxilla. The color scale
lower deviations and a homogenous distribution with (Fig 9) shows the deviations in Figures 6, 7, and 8. The
above-average values in the molar region and the mean deviations in all groups are displayed in the
gingival sulcus. The maximum deviation averaged Table.
423 mm. Deviations in the mandible were significantly The Kolmogorov-Smirnov test showed that the
lower than in the maxilla. overall deviations were significantly different for every
In group 3, the deviations between the virtual models method.
rendered from repeated model scanning with the In Figure 10, the boxplot diagram shows the
D250 were compared. Models deviated on average distribution of the deviations for all methods.

American Journal of Orthodontics and Dentofacial Orthopedics September 2013  Vol 144  Issue 3
476 Fl€
ugge et al

the steep-angled anterior surface might require

Table. Mean deviation of each method (mm)
additional scans from different angles, as Mehl et al27
iTero intraoral iTero extraoral D250 have already suggested for steep areas.
Mean deviation 50 25 10 The imprecise digitization of the molar areas with the
iTero in extraoral use was more pronounced for intraoral
use. This might be caused by the complex angled
DISCUSSION surfaces of the molars and the undercut surfaces of
The reason for the use of intraoral digital impression the neighboring teeth. This theory is supported by
systems is adequate accuracy and precision compared Rudolph et al,15 who used different methods for
with conventional techniques and extraoral digitization digitization of an extraoral reference model to show
of stone casts. Ender and Mehl18 defined accuracy as a that tooth shape was a dominating factor for precision
deviation from the original object and precision as the and that large deviations occurred in areas with strong
accuracy of repeated measurements. changes of curvature. The precision of extraoral model
The precision of intraoral scanning was evaluated in scanning with the D250 was not lower in areas of high
this study and compared with extraoral digitization of curvature and undercuts. The continuous image
stone casts with the iTero and a model scanner (D250). acquisition with laser planes captures all areas of the
The accuracy of the different work flows to create a vir- model precisely, except for interdental spaces, that
tual model was not our objective. The systematic errors accounted for the overall imprecision. The deviations
caused by impression taking and model manufacturing of the group 3 models were on average 10 mm. This
are neither included nor relevant for the present data. agrees with the study of Persson et al.28
Polyether-based stone casts served as an extraoral Despite the identical scanning protocol, the precision
reference. All scans were conducted with the same for extraoral scanning with the iTero (25 mm) was higher
model. Scanning with the model scanner D250 has a than for intraoral scanning with the iTero (50 mm). This
different image acquisition technique compared with might be due to patient movement, limited intraoral
intraoral scanners. The model is continuously captured space, intraoral humidity, and saliva flow. High
with the projection of laser planes and the recording of deviations in intraoral scans of the molar areas indicate
their reflections. With all intraoral scanning techniques, that patient-related factors had a strong influence on
image acquisition is done incrementally. The iTero scanning quality. In comparison, the reduced values
system produces single images of every tooth, which obtained with the extraoral iTero scanning might result
are assembled for a virtual model of the whole jaw. from greater freedom of placement of the scanning
This process, called stitching, might produce a system- wand next to the model teeth.
atic error.26 Mehl et al27 found lower accuracy of Numerous in-vitro studies have shown that prerequi-
quadrant digitization compared to single-tooth sites for clinical use of intraoral and extraoral scanning
digitization with the CEREC system. Because the stitch- devices are met.9,12,15-17 However, the precision of
ing algorithm of the iTero system is unknown, its scanning devices under intraoral conditions has not
contribution to the error in precision cannot be been documented to date. This study shows that the
explained. However, lower precision of the iTero iTero system in vivo and ex vivo can be used to
compared with the D250 was observed in this study. create virtual models for diagnostics and treatment
Loss of information at model edges, especially in the planning in orthodontics. To manufacture orthodontic
maxilla, was observed with the iTero system. This appliances on the basis of the virtual models, not only
resulted in high deviation values. According to the the scanning process but also the production process
scanning protocol, a fixed number of pictures is acquired must be considered. Computer-aided design and
of every tooth. The image section of the camera covers computer-aided manufacturing on the basis of virtual
the tooth and, depending on its anatomy, a variable models created with data from the iTero intraoral scans
portion of the gingiva. The scans show that the image are accompanied by inaccuracies, especially when the
acquisition of the marginal gingiva could not be depiction of facial surfaces of anterior teeth and molars
precisely reproduced in the maxilla. An extended or marginal soft tissues is important for the appliance
protocol, resulting in a longer scanning time, might be (eg, aligners, customized brackets). Extraoral scanning
necessary to obtain complete information. techniques show higher precision and therefore allow
The facial surfaces of the anterior teeth were higher accuracy of applicances built with computer-
imprecisely captured with the iTero in intraoral and aided design and computer-aided manufacturing. How-
extraoral digitizations (Figs 6 and 7). Although they ever, the inaccuracies of the impression must be added
appear to be easily accessible with the scanning wand, to all laser scanning data acquired ex vivo.

September 2013  Vol 144  Issue 3 American Journal of Orthodontics and Dentofacial Orthopedics
ugge et al 477

Fig 10. Boxplot of the deviations for every method (outliers are hidden).

We showed that intraoral scanning with the iTero was features, extraoral scanning has the highest precision.
less precise than extraoral scanning and digitization with For treatment planning and manufacturing of tooth-
the D250, which is still the most precise digitization supported appliances, virtual models created with the
method currently available. The precision of the intraoral iTero can be used.
iTero scan is similar to the values documented in the
literature with conventional polyether impressions REFERENCES
(61.3 6 17.9 mm) for reproduction of the intraoral
1. Morris JB. CAD/CAM options in dental implant treatment
situation. The in-vitro precision for the CEREC
planning. J Calif Dent Assoc 2010;38:333-6.
(30.9 6 7.1 mm) was comparable with that of the 2. Patel N. Integrating three-dimensional digital technologies for
iTero under extraoral conditions (25 mm) in our study.18 comprehensive implant dentistry. J Am Dent Assoc 2010;
The in-vitro precision of the Lava-COS system 141(Supp 2):20-4S.
(60.1 6 31.3 mm) was lower than the in-vitro and 3. Boldt F, Weinzierl C, Hertrich K, Hirschfelder U. Comparison of
the spatial landmark scatter of various 3D digitalization methods.
in-vivo precision of the iTero.18 With CEREC and iTero,
J Orofac Orthop 2009;70:247-63.
the images are recorded in a static position relative to 4. Asquith JA, McIntyre GT. Dental arch relationships on three-
the tooth, whereas the Lava COS captures images while dimensional digital study models and conventional plaster study
the scanning wand is constantly moving; this might models for patients with unilateral cleft lip and palate. Cleft Palate
result in lower precision for this technique. Craniofac J 2012;49:530-4.
5. Sjogren AP, Lindgren JE, Huggare JA. Orthodontic study cast
Our results indicate that the positions of teeth and
analysis—reproducibility of recordings and agreement between
their surfaces can be reproduced in a virtual model, conventional and 3D virtual measurements. J Digit Imaging
but soft-tissue reproduction was imprecise. 2010;23:482-92.
The image acquisition technique of the iTero does 6. Mullen SR, Martin CA, Ngan P, Gladwin M. Accuracy of space
not require the application of a scanning powder. The analysis with emodels and plaster models. Am J Orthod Dentofacial
Orthop 2007;132:346-52.
precision of the iTero might be higher compared with
7. Metzger MC, Hohlweg-Majert B, Schwarz U, Teschner M,
the CEREC and the Lava-COS, since powder application, Hammer B, Schmelzeisen R. Manufacturing splints for orthog-
necessary for these systems, produces a layer of variable nathic surgery using a three-dimensional printer. Oral Surg Oral
thickness.23 Med Oral Pathol Oral Radiol Endod 2008;105(2):e1-7.
8. Persson AS, Oden A, Andersson M, Sandborgh-Englund G.
Digitization of simulated clinical dental impressions: virtual
three-dimensional analysis of exactness. Dent Mater 2009;25:
Intraoral scanning with the iTero is less precise than 929-36.
model scanning with it. Therefore, patient-related 9. Dalstra M, Melsen B. From alginate impressions to digital virtual
models: accuracy and reproducibility. J Orthod 2009;36:36-41.
factors influence the scanning process. Scanning of the
10. Chandran DT, Jagger DC, Jagger RG, Barbour ME. Two- and
maxilla is less accurate than scanning of the mandible. three-dimensional accuracy of dental impression materials: effects
An extended scanning protocol might improve the of storage time and moisture contamination. Biomed Mater Eng
scanning results in the maxilla. Because of its technical 2010;20:243-9.

American Journal of Orthodontics and Dentofacial Orthopedics September 2013  Vol 144  Issue 3
478 Fl€
ugge et al

11. Luthardt RG, Walter MH, Quaas S, Koch R, Rudolph H. Comparison and their constituent measurements. Am J Orthod Dentofacial
of the three-dimensional correctness of impression techniques: a Orthop 2006;129:794-803.
randomized controlled trial. Quintessence Int 2010;41:845-53. 20. Beuer F, Schweiger J, Edelhoff D. Digital dentistry: an overview of
12. Luthardt RG, K€ uhmstedt P, Walter MH. A new method for the recent developments for CAD/CAM generated restorations. Br Dent
computer-aided evaluation of three-dimensional changes in J 2008;204:505-11.
gypsum materials. Dent Mater 2003;19:19-24. 21. Al Mortadi N, Eggbeer D, Lewis J, Williams RJ. CAD/CAM/AM
13. Kurbad A. Impression-free production techniques. Int J Comput applications in the manufacture of dental appliances. Am J Orthod
Dent 2011;14:59-66. Dentofacial Orthop 2012;142:727-33.
14. Kachalia PR, Geissberger MJ. Dentistry a la carte: in-office 22. Hajeer MY, Millett DT, Ayoub AF, Siebert JP. Applications of 3D
CAD/CAM technology. J Calif Dent Assoc 2010;38:323-30. imaging in orthodontics: part II. J Orthod 2004;31:154-62.
15. Rudolph H, Luthardt R, Walter M. Computer-aided analysis of the 23. Meyer BJ, Mormann WH, Lutz F. Optimization of the powder
influence of digitizing and surfacing on the accuracy in dental application in the Cerec method with environment-friendly pro-
CAD/CAM technology. Comput Biol Med 2007;37:579-87. pellant systems. Schweiz Monatsschr Zahnmed 1990;100:1462-8.
16. Del Corso M, Aba G, Vazquez L, Dargaud J, Dohan Ehrenfest DM. 24. Schlager S. Morpho: calculations and visualizations related to
Optical three-dimensional scanning acquisition of the position of geometric morphometrics. 2012.
osseointegrated implants: an in vitro study to determine method 25. R Development Core Team R. R: a language and environment for
accuracy and operational feasibility. Clin Implant Dent Relat Res statistical computing. Vienna, Austria: R Foundation for Statistical
2009;11:214-21. Computing; 2012.
17. DeLong R, Heinzen M, Hodges JS, Ko CC, Douglas WH. Accuracy of 26. Galovska M, Petz M, Tutsch R. Unsicherheit bei der datenfusion
a system for creating 3D computer models of dental arches. J Dent dimensioneller messungen. tm-Technisches Messen 2012;79:
Res 2003;82:438-42. 238-45.
18. Ender A, Mehl A. Full arch scans: conventional versus digital 27. Mehl A, Ender A, Mormann W, Attin T. Accuracy testing of a new
impressions—an in-vitro study. Int J Comput Dent 2011;14:11-21. intraoral 3D camera. Int J Comput Dent 2009;12:11-28.
19. Stevens DR, Flores-Mir C, Nebbe B, Raboud DW, Heo G, Major PW. 28. Persson A, Andersson M, Oden A, Sandborgh-Englund G. A
Validity, reliability, and reproducibility of plaster vs digital study three-dimensional evaluation of a laser scanner and a touch-
models: comparison of peer assessment rating and Bolton analysis probe scanner. J Prosthet Dent 2006;95:194-200.

September 2013  Vol 144  Issue 3 American Journal of Orthodontics and Dentofacial Orthopedics