Available via license: CC BY 4.0
Content may be subject to copyright.
Page 1/17
Facial three-dimensional surface imaging: reliability
and validity of handheld structured light scanners
and a static stereophotogrammetry system
J.A.M. Schipper ( j.a.m.schipper@umcg.nl )
University Medical Center Groningen
B.J. Merema
University Medical Center Groningen
M.H.J. Hollander
University Medical Center Groningen
F.K.L. Spijkervet
University Medical Center Groningen
P.U. Dijkstra
University Medical Center Groningen
J. Jansma
University Medical Center Groningen
R. H. Schepers
University Medical Center Groningen
J. Kraeima
University Medical Center Groningen
Article
Keywords:
Posted Date: August 28th, 2023
DOI: https://doi.org/10.21203/rs.3.rs-3212586/v1
License: This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License
Page 2/17
Abstract
Introduction
Several new systems of three-dimensional (3D) surface imaging of the face have become available to
assess changes following orthognathic or facial surgery. Before they can be implemented into practice
their reliability and validity must be established. Our aim was therefore to study 3dMD
(stereophotogrammetry), Artec Eva and Artec Space Spider (both structured light scanners) regarding
intra- and inter-system reliability and validity.
Material and Methods
Intra- and inter-system reliability expressed in root mean square (RMS) distance was determined by
scanning multiple times a mannequin’s head and faces of healthy volunteers. Validity was determined by
comparing linear measurements of scans with known distances of a 3D printed model. Post-processing
errors were calculated.
Results
Intra-system reliability of mannequin’s head was best for the Artec Space Spider (0.04 mm Spider; 0.07
mm 3dMD; 0.08 mm Eva;). Inter-system reliability of the mannequin’s head showed least differences
between Artec Space Spider and Artec Eva. Intra-system reliability of human subjects was best for the
Artec Space Spider (0.15 mm Spider; 0.20 mm Eva; 0.23 mm 3dMD). Inter-system reliability of human
subjects showed least differences between Artec Eva and Artec Space Spider. Validity of linear
measurements was most accurate in the Artec Space Spider. Post-processing error was 0.01mm for all
systems.
Conclusion
The Artec Space Spider was the most reliable and valid scanning system.
Introduction
Three-dimensional (3D) surface imaging of the face is used for planning and evaluation of orthognathic,
facial surgery and for research purposes1. The 3dMD (3dMD Inc., Atlanta, GA, USA) is the most widely
used scanning system using stereophotogrammetry of which the reliability and validity is reported in
literature2,3. However, disadvantages are the system’s immobility and that certain areas of the face are
not imaged completely due to a xed camera orientation and focus point of the cameras. An advantage
is that the modular system of the 3dMD can have several set-ups enabling it to scan the whole body or
parts of it. But the modular system may lead to different measurement results due to different number of
pods and different lenses.
Several handheld scanners using structured light scanning are gaining popularity and results have been
compared to those of the 3dMD system 4–6. However, these structured light scanners use a sequence of
pictures that form the 3D model instead of an individual 3D picture, whereby accuracy may be lost during
Page 3/17
post-processing. Before clinical application of these new hand-held structured light scanners reliability
and validity needs to be determined.
The Artec Eva (Artec Group Inc., Luxembourg, Luxembourg) has been used in multiple studies, showing
poorer reliability then the 3dMD system 4. According to manufacturer specications, the Artec Space
Spider (Artec Group Inc., Luxembourg, Luxembourg) has a better validity than the Eva system
(www.artec3d.com). This manufacturer determined validity was obtained by scanning “scale ball bars”
with known distances between balls. However, no study compared reliability and validity of the Artec
Space Spider to the 3dMD system in scanning the facial region. The Artec Eva has a larger working
distance than the Artec Space Spider making the Eva system more useful for scanning larger areas such
as the whole body. The manufacturer recommends the Artec Eva as scanner of choice to image the face.
The Artec Space Spider has a narrow eld of view which leads to many small images stitched together to
create a 3D facial model. This procedure may introduce processing errors when scanning larger areas.
Since most literature is available about the 3dMD, we selected two handheld structured light scanners
with a different working distance and different reported validity to compare with the 3dMD. The aim of
this study is therefore to determine the reliability and validity of 3D surface imaging systems Artec Eva,
Artec Space Spider and 3dMD for the facial region by calculating intra-system, inter-system reliability and
validity of linear measurements.
Material and methods
The study was approved by the medical ethics committee of the University of Groningen and University
Medical Centre Groningen (study number METC2021/476). The study protocol was in accordance with
institutional guidelines and the Declaration of Helsinki. Written informed consent was obtained from all
the participants prior to the study. COSMIN taxonomy of measurement properties was used for
terminology and denitions of reliability and validity (www.cosmin.nl). Subjects were healthy employees
and students recruited from the department of oral and maxillofacial surgery of the University Medical
Center Groningen (UMCG). An email was sent to the employees and students to ask for their participation
in this study. Volunteers were excluded from the study if there was a severe deformity of the face if they
had excessive facial hair or if they ever experienced epilepsy (because of the bright ashing light of the
hand-held scanners). Volunteers were scanned with the static 3dMD system and hand-held scanners
Artec Eva and Artec Space Spider. To ensure the natural head position, volunteers were asked to look into
the mirror, gently bite in maximum intercuspation, swallow before the scanning, relax their lips and keep
their eyes open 7.
Technical comparison of the scanning / scanner systems
The working distance of Artec Eva is smallest (0.2 to 0.3 m). The acquisition time and the processing
time of 3dMD are smallest, 1.5 ms and immediate respectively. The manufacturer reported accuracy
i.e.
validity is highest in the Artec Space Spider (0.1mm). The models produced by the Artec Space Spider
capture most details (Table1).
Page 4/17
Table 1. Overview of scanner systems
3dMDface system Artec Eva Artec Space Spider
3D model example
Modality Stereophotogrammetry Structured light Structured light
Handheld
Working distance Not applicable Yes
(0.2m - 0.3m)
Yes
(0.4m – 1.0m)
Acquisition time 1.5ms ~ 20 s ~ 60s
Processing time Immediate
reconstruction Up to 10 minutes Up to 30 minutes
Accuracy (reported by
manufacturer) 0.2mm 0.1 mm 0.05 mm
Resolution 0.5 mm 0.5 mm 0.1 mm
Cost >70 000 USD >15 000 USD >25 000 USD
Data processing and analysis
Data obtained using Artec Eva and Artec Space Spider were processed using Artec Studio 14 Professional
software (Artec Group, Luxembourg, Luxembourg) to acquire 3D models (Fig.1). 3dMD images used
Vultus software (3dMD, Inc, Atlanta, GA, USA). As Artec Artec Eva and Artec Space Spider need additional
steps to acquire a 3D model from the several frames, a standardized protocol for the processing in Artec
Studio was used, following the manufacturer’s website 8. In 3dMD Vultus software, surface-based
registration based on the iterative closest point method was performed for all models to match the scans.
After images were manually aligned in 3dMD Vultus software, a specic T-shaped area of the forehead
(Fig.1) and dorsum of the nose was marked on each scan as surface for the registration. The quality of
the registration was determined in the Vultus software and expressed as registration root mean square
(RMS) error. The quality of the matching process was also determined based on visual inspection. Scans
were then cut with 3-matic software (Materialise, Leuven, Belgium) by selecting planes formed by the
tragus-forehead and tragus-menthon (Fig.1). Thereafter scans were imported in CloudCompare
(CloudCompare, version 2.11 alpha) and the cloud-to-mesh distance with automatic octree level was
Page 5/17
computed to produce signed distances (Fig.1). In this mode, for each point of the compared cloud
CloudCompare will simply search the nearest triangle in the reference mesh. RMS was calculated from
these signed distances using RStudio (version 2021.09.2 + 382).
Mannequin’s head intra-system reliability
For intra-system reliability of the scanners without interference of facial variability, a mannequin’s head
was scanned three times with each system (Fig. 2). The RMS distances between the 3 scan pairs (1st
scan vs 2nd, 1st vs 3rd, 2nd vs 3rd ) of each system were calculated.
Mannequin’s head inter-system reliability
Distance maps were calculated after registration in 3dMD Vultus software using histograms with
identical color corresponding differences. Data acquired for the intra-system reliability were analyzed
again. For inter-system reliability analysis the RMS distances between 1st scans of each system, between
the 2nd scans of each system and the 3rd scans of each system were calculated (thus 1st scan 3dMD vs
1st scan Artec Eva, 1st scan 3dMD vs 1st scan Artec Spider, and 1st scan Artec Spider vs 1st scan Artec
Eva etc.).
Intra-system reliability in 3 volunteers
To obtain the intra-system reliability in humans 3 volunteers were scanned 6 times with each system
(Fig.2). RMS distance for all 15 scan pairs (1st scan vs 2nd, 1st vs 3rd, 1st vs 4th, 1st vs 5th, 1st vs 6th,
2nd vs 3rd, 2nd vs 4th, 2nd vs 5th, 2nd vs 6th, 3rd vs 4th, 3rd vs 5th, 3rd vs 6th, 4th vs 5th, 4th vs 6th, 5th
vs 6th ) for each volunteer for each system were calculated.
Inter-system reliability in 16 volunteers
To obtain inter-system reliability in humans 16 volunteers were scanned with each imaging system once
(Fig.2). RMS distances were calculated for Artec Eva versus 3dMD, Artec Space Spider versus 3dMD and
Artec Eva versus Artec Space Spider for each participant. Mean RMS distances were calculated for these
pairs of the participants combined.
Validity using a reverse engineered mannequin’s head as
gold standard
Since the precise surface dimensions of the mannequin’s face were unknown, we reverse engineered the
mannequin’s face. The face was scanned using the Artec Space Spider and cylinders with known linear
dimensions in multiple directions were modelled in the scan of the mannequin’s head using 3-matic
software (Materialise, Leuven, Belgium). A 3D printer (PA12 material, EOS SLS printer, Oceanz, Ede, The
Netherlands) was then used to print this reference model with known dimensions (Fig. 3). To compensate
for potential errors in the 3D printing process, precise calipers (micrometers) with accuracy of 0.01 mm
were used to take linear measurements using 2 observers with 3 repeated measurements of the length of
the cylinders of the 3D printed mannequin model to act as gold standard. The 3D printed face was then
scanned three times consecutively with the 3dMD, Artec Eva and Artec Space Spider system. Linear
Page 6/17
measurements of these 3D scans were then taken from these cylinders in 3-matic software (Materialise,
Leuven, Belgium). The difference between the gold standard linear measurements and the 3D scanned
measurements were used as validity measures. RMS distances were calculated for the 3dMD, Artec Eva
and Artec Space Spider system compared to the reference model (the mesh le that was 3D printed).
Post-processing error
Post-processing error was recorded by repeating the analysis process 5 times for one scan pair.
Statistical analysis
All statistical analyses were performed using RStudio (version 2021.09.2 + 382). For the mannequin intra-
and inter-system reliability descriptive statistics were provided, no statistical testing was performed due
to the limited sample size of 1 mannequin. Continuous variables were denoted as mean and standard
deviation (SD). Normal distribution or skewness was assessed by visual inspection of histograms and
statistically tested by the Shapiro-Wilk test. One-way repeated measures analysis of variance (ANOVA)
with post hoc Tukey’s honestly signicant difference test was performed for the inter system reliability on
human subjects. The Friedman test was performed as nonparametric test when the conditions for
parametric testing were not met.
Results
Mannequin’s head intra-system reliability
Intra-system reliability of the Artec Space Spider was best (RMS 0.04mm). 3dMD and Artec Eva showed a
similar reliability (Table 2a). Registration RMS error was in the range of 0.02-0.04mm.
Table 2a. Results intra system reliability study on mannequin’s head, scanned 3 times with each system
3dMD
(RMS in mm±SD)
Artec Eva
(RMS in mm ±SD)
Artec Space Spider
(RMS in mm ±SD)
Mean difference±SD 0.07±0.01 0.08±0.02 0.04±0.01
Registration error 0.02 0.04 0.02
RMS: root mean square. The RMSrepresents the mean difference between the 1st scan vs 2nd, 1st vs 3rd,
2nd vs 3rd.
Table 2b. Results of inter system reliability study on mannequin’s head, scanned 3 times with each
system
Page 7/17
3dMD vs. Artec Eva
(RMS in mm ±SD)
3dMD vs. Artec Space
Spider
(RMS in mm ±SD)
Artec Space Spider vs.
Artec Eva
(RMS in mm ±SD)
Distance map
1st scan 0.22±0.20 0.17±0.15 0.08±0.15
2nd scan 0.19±0.18 0.19±0.19 0.15±0.22
3rd scan 0.21±0.17 0.16±0.14 0.12±0.16
Mean
difference±SD 0.21±0.01 0.17±0.03 0.12±0.01
Registration
error 0.07 0.06 0.06
RMS: root mean square. The RMS representsthe difference between the 3dMD vs Artec Eva, 3dMD vs
Artec Space Spider and Artec Space Spider vs Artec Eva system. Fine lines are visible on the head (arrow).
These lines were cut into the mannequin for a previous study.
Mannequin’s head inter-system reliability
Visual inspection of the distance maps between systems illustrated that 3dMD was less reliable than
Artec Eva and Artec Space Spider and showed especially differences in the following regions: under the
nose, under the lip, at the side of the face (Table 2b). Distance maps showed more positive difference of
the Artec systems compared to the 3dMD, especially at the sides of the face. Clinically this means that
the Artec systems over-estimates or the 3dMD under-represents the true shape of the side of the face. In
the distance map of Artec Eva vs. Artec Space Spider ne lines can be seen that were cuts made into this
mannequin for a previous study. These cuts were better captured by the Artec Space Spider than the Artec
Eva. Registration RMS error was 0.06mm for Artec Eva vs Artec Space Spider, 0.08mm for Artec Eva vs.
3dMD and 0.06mm for Artec Space Spider vs 3dMD.
Concluding inter-system reliability between Artec Space Spider and Artec Eva was best.
Human subjects
Page 8/17
Sixteen volunteers were scanned, but 14 scans were included in the statistical analyses since 2
volunteers had facial hair that produced errors on the surface models. For the intra-system reliability
measurements 3 volunteers were randomly selected to obtain extra scans. All 14 scans were used for the
inter-system reliability.
Intra-system reliability in 3 volunteers
Shapiro-Wilk test showed no signicant deviation from the normal distribution. Intra-system
reproducibility did not differ signicantly between the systems (repeated measures one-way ANOVA, p =
0.498). The mean differences, in RMS, were smallest for Artec Space Spider (0.15 ± 0.02 mm) (Table3).
Registration RMS error was 0.07mm for 3dMD, 0.09mm for Artec Eva and 0.07mm for Artec Space
Spider.
Table 3
Intra-system reliability of 3 human subjects, scanned 6 times with each system
3dMD
(RMS in mm)
Artec Eva
(RMS in mm)
Artec Space Spider
(RMS in mm)
Person 1
(15 scan pairs)
0.17 ± 0.03 0.26 ± 0.09 0.13 ± 0.05
Person 2
(15 scan pairs)
0.33 ± 0.12 0.16 ± 0.08 0.14 ± 0.03
Person 3
(15 scan pairs)
0.18 ± 0.03 0.18 ± 0.08 0.18 ± 0.04
Mean difference ± SD 0.23 ± 0.08 0.20 ± 0.05 0.15 ± 0.02
Registration error 0.10 0.06 0.06
RMS: Root mean square. RMS represents the mean difference between the scans for every person.
Six scans of every person were compared in every possible combination which resulted in 15 scan
pairs (1st scan vs 2nd, 1st vs 3rd, 1st vs 4th, 1st vs 5th, 1st vs 6th, 2nd vs 3rd, 2nd vs 4th, 2nd vs 5th,
2nd vs 6th, 3rd vs 4th, 3rd vs 5th, 3rd vs 6th, 4th vs 5th, 4th vs 6th, 5th vs 6th ). Intra-system
reproducibility did not differ signicantly between the systems (repeated measures one-way ANOVA,
p = 0.498)
Inter-system reliability in 14 volunteers
Shapiro-Wilk test showed deviation from the normal distribution for the 3dMD vs Eva (p = 0.01), 3dMD vs
Spider (p = 0.02) but not for Spider vs Eva (p = 0.20), therefore Friedman test was used. Inter-system
reliability differed signicantly between the three comparisons, Artec Eva-3dMD, Artec Space Spider-
3dMD and Artec Eva-Artec Space Spider (Friedman test; χ2 = 22.29; p < 0.01). Post-hoc Conover testing
showed least difference of Spider vs Eva compared to 3dMD vs Eva (p < 0.01; 0.26 ± 0.06 mm vs 0.39 ±
Page 9/17
0.11 mm, respectively), least difference of Spider vs Eva compared to 3dMD vs Spider (p < 0.01; 0.26 ±
0.06 mm vs 0.45 ± 0.11 mm, respectively) but no difference between 3dMD vs Eva and 3dMD vs Spider
(p = 0.80; 0.39 ± 0.11 mm vs 0.45 ± 0.11 mm, respectively). Comparisons of 14 volunteers showed the
best inter-system reliability between Artec Eva and Artec Space Spider (Table 4). Registration RMS error
was 0.12 mm for Artec Eva vs. Artec Space Spider, 0.14mm for Artec Eva vs. 3dMD, 0.13mm for Artec
Space Spider vs. 3dMD.
Table 4
Inter-system reliability of 14 human subjects, scanned 1 time with each system
3dMD vs.
Artec Eva
(RMS in
mm ± SD)
3dMD vs. Artec Space Spider
(RMS in mm ± SD) Artec Space Spider vs. Artec Eva
(RMS in mm ± SD)
10.31 ± 0.31 0.33 ± 0.52 0.26 ± 0.24
20.69 ± 0.57 0.47 ± 0.53 0.21 ± 0.19
30.53 ± 0.64 0.41 ± 0.75 0.31 ± 0.36
40.27 ± 0.35 0.48 ± 0.95 0.24 ± 0.23
50.34 ± 0.25 0.38 ± 0.26 0.10 ± 0.14
60.35 ± 0.39 0.37 ± 0.55 0.21 ± 0.27
70.29 ± 0.75 0.34 ± 0.83 0.29 ± 0.25
80.32 ± 0.34 0.50 ± 0.88 0.24 ± 0.22
90.45 ± 0.54 0.75 ± 0.17 0.31 ± 0.27
10 0.42 ± 0.39 0.44 ± 0.54 0.26 ± 0.23
11 0.32 ± 0.33 0.40 ± 0.74 0.26 ± 0.24
12 0.45 ± 0.41 0.44 ± 0.74 0.30 ± 0.31
13 0.35 ± 0.35 0.34 ± 0.62 0.34 ± 0.32
14 0.35 ± 0.35 0.59 ± 1.00 0.34 ± 0.28
Mean
difference ±
SD
0.39 ± 0.11 0.45 ± 0.11 0.26 ± 0.06
Registration
error 0.14 0.13 0.12
RMS: Root mean square. RMS represents the mean difference between the scans of the different
systems. Friedman test and post-hoc Conover testing showed signicant differences between 3dMD
vs Eva and 3dMD vs Spider (p < 0.01), 3dMD vs Spider and Spider vs Eva (p < 0.01), but no difference
between 3dMD vs Eva and 3dMD vs Spider (p = 0.80).
Page 10/17
Validity using reverse engineered mannequin’s head as gold
standard
Linear measurements digitally performed on the 3D scans compared to the gold standard showed that
the Artec Space Spider had the best validity (Table 5). The Artec Eva showed consistently higher
measurements, while the 3dMD showed consistently lower measurements. When comparing the 3D
printed model to the scans of the three systems, the Artec Space Spider was also most accurate in
scanning the geometry showing the lowest RMS difference between these meshes. Distance maps
showed that the scans of the three systems mostly differed in the lower part of the face.
Table 5
Reverse engineered mannequin’s head with cylinders for gold standard
3dMD vs.
mannequin
model
Artec Eva vs.
mannequin model Artec Space Spider vs.
mannequin model
Distance maps
Linear measurements error
(deviation in mm ± SD) 0.30 ± 0.07 0.30 ± 0.10 0.03 ± 0.02
Mean difference (RMS in mm
± SD) 0.38 ± 0.06 0.36 ± 0.07 0.31 ± 0.03
Registration error 0.21 0.39 0.10
Post-processing error
After repeating all described processing steps ve times for one scan pair, a standard deviation of 0.01
mm was found for 3dMD, 0.01 mm for Artec Eva and 0.01 mm for Artec Space Spider.
Discussion
Three-dimensional surface imaging has proven itself as a valid and reliable imaging modality for
evaluation of several surgical treatment results of the face. New handheld scanning systems are being
released which advertise even higher accuracy than previously widely used static systems. Reliability
studies of these new scanners are needed before they can be safely used in clinical practice and
research.
Page 11/17
Intra-system and inter-system reliability in this study was established by scanning both healthy
volunteers and a mannequin’s head. We found that the Artec Space Spider had a better intra-system
reliability for both volunteers and the phantom head. The Artec Space Spider captures more details, since
little wrinkles and small protuberances of the face can be visualized very well. Comparisons between the
Artec Space Spider and Artec Eva showed small differences in reliability. However, compared to these
handheld scanners the 3dMD system introduced more inaccuracy as illustrated by the distance maps
especially under the nose, under the chin and at the side of the face. These differences could possibly be
explained by the xed orientation of the 3dMD system. The Artec handheld systems can scan in several
angles around the face to capture iterative meshes that are stitched together to produce a complete 3D
model. Therefore, geometry underneath the nose or chin can be imaged more completely using handheld
systems. Because of the narrow eld of view of the Artec Space Spider, a lot of iterative meshes must be
captured and stitched together to create a 3D model. This process can hypothetically introduce a
processing error when scanning large areas. However, since we found an overall high reliability and
validity of the Artec Space Spider when scanning an area of the size of the face, we regard the relevance
of this error low.
Moreover, since errors were all below 0.5mm, the clinical relevance of these differences are very limited.
Other properties such as ease-of-use or time needed to generate a 3D model could dictate the choice for a
specic system. However, when measuring volume by comparing larger areas of surface scans of the
face, these small differences add up to more inaccurate volume measurements. Therefore, it is
recommended for pre- and post-surgery volume measurements, to use a system with the smallest errors
such as the Artec Space Spider.
In previous studies, 3dMD was used as reference to compare other scanners with 6. Since reliability of
other scanners is better than 3dMD in this study, we believe that 3dMD should not act as a gold standard
anymore. We 3D printed a mannequin phantom head with cylinders. Since 3D printing introduces errors,
we took linear measurements of the cylinders, in multiple directions on the mannequin’s head, to
operationalize this as gold standard. These measurements showed that Artec Space Spider is more valid
in capturing the geometry of the face. Based on the linear measurements error, it seems that the Artec Eva
was overestimating the volume of the face and the 3dMD was underestimating the volume of the face.
Most studies analyzed reliability and validity of the 3dMD stereophotogrammetry system 2,3,9. A recent
study scanning dental casts showed that the Artec Space Spider had high reliability, but low validity
(0.4mm) 10. However, the authors attributed the error to the automatic reconstruction function of the
software, a function we did not use. Artec Space Spider showed one of the best validity measures when
compared to Primescan, Trios, Pritiface and iPhone when used to scan nasal, orbital and auricular
models that were manufactured using stereolithography11. Artec Space Spider was found to be most
valid compared to Artec Eva, Vectra H1, Bellus and SNAP when used to scan plaster statues with balls
attached of which linear measurements were used for comparison12. In the peri-orbital region an excellent
reliability of 0.1–0.2 mm of the Artec Space Spider was found 13.
Page 12/17
In this study we showed that all scanners produce clinically acceptable results, with errors below 0.3mm
for the intra-system reliability. Variation in facial expression is known to introduce an error when
comparing multiple facial 3D surface scans, which explains the differences between the human
volunteers and the phantom head14. The substantial difference between the phantom head and the face
RMS distances shows that the variability of the expression of the face introduces more error than the
scanning systems themselves. This nding highlights that standardization of facial expression during
scanning could be even more important than scanning system accuracy. For clinical use, certain
properties can differentiate their usefulness for specic applications. The Artec Space Spider is especially
useful for acquiring scans of objects with complex geometry, since it showed the best validity and
captured most detail. Artec Eva and Artec Space Spider are mobile scanners, a property which makes it
especially useful when the scanner needs to be used in multiple sites, such as the operating room or
multiple clinical centers. In addition to this, the storage space of these mobile scanners is substantially
less than the static scanning system 3dMD, which needs a designated room for its setup. However,
training and experience is required for the mobile scanners to produce an acceptable quality of the scans
since slow and homogenous movements are required during the recording. The 3dMD obtains surface
models with a single ash which reduces the risk of movement artefacts during scanning and provides
fast acquisition times.
One of the strengths of our study is that we included both intra- and inter-system reliability of a
mannequin’s head and human subjects. This facilitates proper assessment of the reliability of both a
rigid object with the geometry of the face and the face of a human subject. We also included validity
analysis of a 3D printed face and linear measurements of cylinders on this 3D print. However, the main
limitation of this study is that for validity measures the true geometry of the human face is not known. It
is therefore impossible to dene a gold standard which facilitates assessment of the validity of capturing
the true complex geometry of the face. We therefore included assessment of a 3D printed face and
compared this with the original printed mesh. 3D printing however also introduces a printing error. We
therefore included assessment of linear measurements with calipers and compared these with the linear
measurements of the 3D scan which showed again that the Artec Space Spider was most valid. Another
limitation is the small sample size in this study. Despite these limitations multiple analyses showed
consistent results: the Artec Space Spider scanning system was most reliable and valid. Future research
on validity of new 3D surface imaging systems should therefore focus on establishing a proper gold
standard for facial scanning.
We provided evidence that three scanning systems in our comparison were reliable and valid and
concluded that the Artec Space Spider was most reliable and valid. Other properties such as ease-of-use
or mobility of the 3D surface imaging systems can dictate which system to choose in practice.
Nonetheless, 3D surface imaging is a promising and radiation-free imaging modality that can be safely
implemented to assess facial surgical outcomes during follow-up.
Declarations
Page 13/17
Author contributions
Schipper. J.A.M.: Writing – Original draft, Conceptualization, Investigation, Formal analysis, Data curation
B.J. Merema: Writing – Review and Editing, Software, Data curation
M.H.J. Hollander: Writing – Review and Editing, Conceptualization
F.K.L. Spijkervet: Writing – Review and Editing, Supervision
P.U. Dijkstra: Writing – Review and Editing, Methodology, Supervision
J. Jansma: Writing – Review and Editing, Conceptualization, Supervision
R.H. Schepers: Writing – Review and Editing, Conceptualization, Supervision
J. Kraeima: Writing – Review and Editing, Conceptualization, Methodology, Software, Supervision
Data availability statement
Data is available at the research databank of the department of Oral and Maxillofacial Surgery in the
University Medical Centre of Groningen. Data can be requested through the rst author, J.A.M. Schipper
(j.a.m.schipper@umcg.nl). It can also be requested through our department administration
+31503613840 (l.kempers@umcg.nl).
No competing interests, no disclosure.
The authors declare that they have no conict of interest. There are no nancial interests in any of the
companies that were mentioned in this manuscript by any of the authors.
Figure consent
The human subject in the gures is the rst author of this manuscript. The rs author provides consent
for publishing the gures.
References
1. Kau, C. H., Richmond, S., Incrapera, A., English, J. & Xia, J. J. Three-dimensional surface acquisition
systems for the study of facial morphology and their application to maxillofacial surgery. Int. J. Med.
Robot. 3, 97–110 (2007).
2. Lübbers, H. T., Medinger, L., Kruse, A., Grätz, K. W. & Matthews, F. Precision and accuracy of the 3dmd
photogrammetric system in craniomaxillofacial application. J. Craniofac. Surg. 21, 763–767 (2010).
Page 14/17
3. Aldridge, K., Boyadjiev, S. A., Capone, G. T., DeLeon, V. B. & Richtsmeier, J. T. Precision and error of
three-dimensional phenotypic measures acquired from 3dMD photogrammetric images. Am. J. Med.
Genet. A 138A, 247–253 (2005).
4. Verhulst, A.
et al.
Three-Dimensional Imaging of the Face: A Comparison between Three Different
Imaging Modalities. Aesthetic Surg. J. 38, 579–585 (2018).
5. Modabber, A.
et al.
Evaluation of the accuracy of a mobile and a stationary system for three-
dimensional facial scanning. J. Cranio-Maxillofacial Surg. 44, 1719–1724 (2016).
. Knoops, P. G. M.
et al.
Comparison of three-dimensional scanner systems for craniomaxillofacial
imaging. J. Plast. Reconstr. Aesthetic Surg. 70, 441–449 (2017).
7. Jakobsone, G., Vuollo, V. & Pirttiniemi, P. Reproducibility of Natural Head Position assessed with
stereophotogrammetry. Orthod. Craniofacial Res. 23, 66–71 (2020).
. Artec3D Website. https://www.artec3d.com/portable-3d-scanners.
9. Metzger, T. E., Kula, K. S., Eckert, G. J. & Ghoneima, A. A. Orthodontic soft-tissue parameters: A
comparison of cone-beam computed tomography and the 3dMD imaging system. Am. J. Orthod.
Dentofac. Orthop. 144, 672–681 (2013).
10. Piedra-Cascón, W., Methani, M. M., Quesada-Olmo, N., Jiménez-Martínez, M. J. & Revilla-León, M.
Scanning accuracy of nondental structured light extraoral scanners compared with that of a dental-
specic scanner. J. Prosthet. Dent. 126, 110–114 (2021).
11. Unkovskiy, A.
et al.
Accuracy of capturing nasal, orbital, and auricular defects with extra- and
intraoral optical scanners and smartphone: An in vitro study. J. Dent. 117, 103916 (2022).
12. Tsuchida, Y., Shiozawa, M., Handa, K., Takahashi, H. & Nikawa, H. Comparison of the accuracy of
different handheld-type scanners in three-dimensional facial image recognition. J. Prosthodont. Res.
(2022) doi:10.2186/jpr.JPR_D_22_00001.
13. Hollander, M. H. J.
et al.
Reproducibility of 3D scanning in the periorbital region. Sci. Rep. 11, (2021).
14. Maal, T. J. J.
et al.
Variation of the face in rest using 3D stereophotogrammetry. Int. J. Oral
Maxillofac. Surg. 40, 1252–1257 (2011).
Figures
Page 15/17
Figure 1
Steps of processing
Page 16/17
Figure 2
Study diagram
Page 17/17
Figure 3
Reverse engineered 3D printed mannequin with known geometry and cylinders