Version of Record: https://www.sciencedirect.

com/science/article/pii/S1761722720300991
Manuscript_799f3ddf0e7bb0f25c0a8c342505c3c2

Original article
Voxel-based superimposition of Cone Beam CT scans for
orthodontic and craniofacial follow-up: overview and clinical
implementation

Gauthier Dot1*, Frédéric Rafflenbeul2, Benjamin Salmon3

1
Université de Paris, AP-HP, Hopital Pitié-Salpétrière, Service d’Odontologie. Institut de
Biomécanique Humaine Georges Charpak (IBHGC), Arts et Métiers ParisTech Paris, France
2
Département d’Orthopédie Dento-Faciale, Faculté Odontologie, Université de Strasbourg, 67000
Strasbourg, France
3
Université de Paris, Service de Médecine Buccodentaire, Hôpital BretonneauF-75018, Paris, France.

Correspondence:
*Gauthier DOT, Université de Paris, AP-HP, Hopital Pitié-Salpétrière, Service d’Odontologie, 47-83
Boulevard de l'Hôpital, 75013 Paris, France. Institut de Biomécanique Humaine Georges Charpak
(IBHGC), Arts et Métiers ParisTech, 151 bd de l'Hôpital, 75013 Paris, France
E-mail : gauthier.dot@ensam.eu
Tél : +33606611627

Abstract
Introduction> The increasing use of three-dimensional (3D) imaging in orthodontics has led
to the development of 3D superimposition techniques. These techniques use stable anatomic
structures as references in order to compare Cone Beam CT (CBCT) scans of the same
subject at different time-points. Three methods have been described in the literature:
landmark-based, surface-based and voxel-based 3D superimpositions.
Objective > This article focuses on the voxel-based approach, which is the most described and
the only one that can be fully automatized. The aim of this paper is to offer clinicians a
practical tutorial on craniofacial voxel-based 3D superimposition.
Material and Methods > We provide an updated overview of the available implementation
methods, describing their methodology, validations, main steps, advantages and drawbacks.
The historical open-source method is the most widespread for research purposes, but takes
around three hours to achieve for an experienced operator. Several commercially-available
software perform superimpositions in a few minutes.
Results > We used two of the available methods to conduct the superimposition process with
three representative clinical cases in order to illustrate the different types of results that can be
obtained.
Conclusions > Commercially-available software provide user-friendly and fully automatized
superimposition methods, allowing clinicians to perform it easily and helping to reduce
human error in image analysis. Still, quantitative evaluation of the results remains the main
challenge of this technique.

Keywords Imaging. Three-Dimensional. Cone-Beam Computed Tomography. Skull Base /

diagnostic imaging. Facial Bones / diagnostic imaging. Image Processing, Computer-Assisted
/ methods

1
© 2020 published by Elsevier. This manuscript is made available under the Elsevier user license
https://www.elsevier.com/open-access/userlicense/1.0/
Introduction

In dentofacial orthopaedics, “superimpositions” are commonly performed on two-dimensional

(2D) radiographs. By using stable anatomical structures from a same patient at different time-
points as references, it allows the evaluation of changes that occurred by growth and/or
treatment in the other structures [1]. Superimposition of before- and after-treatment lateral
skull radiographs is routinely used and is required in orthodontic board examinations around
the world [2,3]. These two-dimensional (2D) superimpositions of radiographs have
nonetheless several limitations. 2D images are projections of three-dimensional (3D)
craniofacial structures, which lead to image distortion and overlapping of bilateral structures
[4]. Moreover, anatomic structures used as stable references have been shown to be less
reliable than osseous implants [1].

When a 3D imaging follow-up is justified, reliable 3D methods are needed in daily clinical
practice to objectively monitor treatment outcomes and growth changes. These methods
require to align on stable anatomic structures Cone Beam CT (CBCT) or CT-Scan images
obtained from the same patient in different timeframes. In the image analysis community, this
alignment is commonly called “image registration”[5]. In this paper, we have chosen to keep
the word “superimposition” which has been commonly used in the orthodontic community for
decades.

In order to superimpose a T2 image on a T1 reference image, the underlying principles of 3D

superimposition are the same as those used in 2D superimposition [6]. First, the same stable
anatomic zones (region of reference) have to be selected in the two images. Then, a rigid 3D
transformation with 6 degrees of freedom (3 rotations and 3 translations) registers the region
of reference from the T2 image on the same structure in the T1 image (figure 1). If the two
stable anatomic zones do not have the same size, for example in a growing patient, a scaling
of the images can be performed [7]. The changes that occurred in the other structures of the
images can then be assessed.

Three methods have been described for 3D superimposition of different craniofacial images
of a same patient: (1) landmark-based, (2) surface-based and (3) voxel-based [8]. As stated by
a recent literature review, validation studies suggest that the three methods provide an
acceptable level of reliability for general superimpositions [9]. However, all included studies
suffered from a lack of methodological quality. Landmark-based 3D superimposition relies on
localization of landmarks on the stable structures. Manual placement of these landmarks
requires well-trained operators in order to be reproducible and is time-consuming [10,11].
Surface-based (or feature-based) 3D superimposition, on the other hand, registers high quality
3D surface models of the stable structures, which can be tedious to obtain and may imply
scanning protocols with higher exposure [8,12].

In this paper, we will focus on voxel-based (or intensity-based) 3D superimposition, which is

the most described and only technique that can be fully automatized. The automatization
process allows clinicians and researchers to perform image analysis in a non-operator
dependent way [8]. The algorithm is based on the maximization of mutual information (MIM)
theory [13], and is suitable for CT-Scan or CBCT images. Selected reference areas are
automatically compared and matched according to their grayscale values. Voxel-based
superimposition for dentofacial applications was introduced by Cevidanes et al. in 2005 [6].
Their first works focused on general superimpositions, using the cranial base as region of

2
reference. Since then, various other papers explored the relevance of this registration
technique to achieve general or regional superimposition[14-17], for adult or growing patients
[7].

The aim of this paper is to offer clinicians a practical tutorial on craniofacial voxel-based 3D
superimposition. We therefore provide an updated overview of the available implementation
methods and illustrate the various types of results that can be obtained with three clinical
cases. Finally, the main challenges of the technique are discussed.

Materials and Methods

Dental and Craniofacial Bionetwork for Image Analysis method

The Dental and Craniofacial Bionetwork for Image Analysis (DCBIA)[18] group developed
the historical procedure to perform voxel-based 3D superimposition of craniofacial structures
[8]. This process relies on a combination of two open-source software. The first one, ITK-
Snap [19], is used for visualization and semi-automatic segmentation of the region of
reference. The second one, 3D Slicer [20] is used to register the images on the
aforementioned structures and to evaluate the results.
To learn how to conduct this method, clinicians and researchers are invited to watch freely-
accessible YouTube video tutorials developed by the research team [18]. A series of about 20
videos explains the handling of the two software and the main steps to perform 3D
superimposition (figure 2, online).
This method is the most widespread for research purposes, and has been validated in various
studies in general or regional superimposition, for adult or growing patients [9]. It is very
efficient and has the advantage to rely only on freely-accessible, easily-modifiable and open-
source software [17]. However, the clinical application of the method remains difficult as it
requires some computing skills, training and time. Indeed, the superimposition of two 3D
time series takes approximately 3 hours for an experienced operator [21].

Commercially-available software methods

In the last ten years, several commercially-available software allowing voxel-based 3D
superimposition were developed. At least 4 software are now available: Dolphin Imaging
(Dolphin Imaging and Management Solutions, Chatswoth, CA, USA), InVivo Dental
(Anatomage, San Jose, CA, USA), Maxilim (Medicim NV, Mechelen, Belgium) and
OnDemand3D (Cybermed, Seoul, Korea)[8]. General superimpositions have been shown to
be reliable for adult patients in Maxilim software [22], and for both adult and growing patients
in Dolphin Imaging software [21,23]. OnDemand3D method has been positively evaluated
both for adults and growing patients in general [24,25] as well as regional [26,27]
superimpositions.
These software are user-friendly, do not require extensive training of the operators and
perform superimpositions in a few seconds or minutes. For example, the OnDemand3D
procedure is the following: (1) import of T1 and T2 scans in Digital Imaging and
Communication of Medicine (DICOM) format (2) approximate manual registration of the T2
image on the T1 image; (3) selection of the region of reference; (4) algorithm launch; (5)
evaluation of results [28].
This process is shown in Additional file Movie S1 (figure 3), which has been recorded in real-
time and illustrates the use of OnDemand3D software to superimpose two maxillary CBCT
images of patient #2 (see further for case details).

3
Selection of the region of reference
-For general superimpositions of non-growing patients, the region of reference to select is the
whole cranial base[6]. For growing patients, this region should be reduced to the anterior
cranial base, which is supposed to have completed growth by age 7 [7,9]. Cevidanes et al.
described this structure as this: “anterior wall of Sella, anterior clinoid processes, planum
sphenoidale, lesser wings of the sphenoid, superior aspect of ethmoid and cribriform plate,
cortical ridges on the medial and superior surfaces of the orbital roofs, and inner cortical layer
of the frontal bones”[7].
-For maxillary regional superimpositions, Ruellas et al. compared, in growing patients, two
regions of reference that excluded the dentoalveolar processes [14]. The two areas showed
similar results and were considered adequate to use. Therefore, the easiest region to use for
clinicians is probably the smallest one, which use the key ridge as a stable structure. This
region is cropped inferiorly at the dentoalveolar processes; superiorly and bilaterally at the
Orbitale points; posteriorly distal to the first molars and anteriorly distal to the canines [14].
-For mandibular regional superimpositions, Nguyen et al. showed that the chin and symphysis
regions should be used as region of reference in growing patients [16]. The lower contour of
the third molar crypt showed a significant level of displacement, and should not be used as
reference [15,16].

Summary of available methods

For each of these methods, we have summarized in table I the respective validation studies,
regions of reference to use, main steps to follow, required training of the operators, average
duration and advantages/drawbacks.

Selection of our clinical cases

We carried out superimpositions through the DCBIA open-source approach or OnDemand3D
software with three representative clinical cases. In this retrospective study, written informed
consent for experimentation and publication was obtained from all patients (and their parents
for minor patients). The scans had been taken for the purpose of their orthodontic treatment,
therefore none of the patients was exposed to any extra radiation for the purpose of this
project.

Results

Patient #1, alveolar bone grafting in cleft lip and palate (DCBIA open-source method)
A 7-year-old patient with unilateral cleft lip and palate got an alveolar bone graft after
orthodontic maxillary expansion. In order to perform a regional maxillary superimposition,
we retrieved pre-expansion T1 CBCT and post-surgery T2 CBCT which were acquired two
years apart. We followed the DCBIA procedure using open source ITKSnap and 3D Slicer
software, using the region of reference previously described in the Material & Methods
section.
Regional superimposition results can be evaluated in ITKSnap software by navigating in the
axial slices. Here, a very good stability and volume of the bone graft can be confirmed (figure
4 and additional file movie S2). Additional evaluation of the results can be made using the 3D
surface models of the T1 and T2 images and transparent overlays in 3D Slicer software
(figure 5).

4
Patient #2, TAD supported maxillary incisor intrusion (OnDemand3D software)
This 13-year-old patient sought treatment for correction of his maxillary front teeth crowding.
Patient showed a skeletal Class II and a typical Class II division 2 malocclusion (figure 6a).
Treatment plan was to decompensate incisor inclination to a Class II division I and to
subsequently correct the Class II with a functional appliance. In order to get a fast intrusion
and proclination of the incisors, a temporary anchorage device (TAD) was placed between
teeth #11 and #21. A coil spring was activated between the TAD and a sectional wire
connecting teeth #11, #21 and #22. This device provided an intrusion force buccally to the
centre of resistance of these teeth. Decompensation was obtained in three months and was
retained before the functional appliance phase (figure 6b).
In order to perform a regional maxillary superimposition, we retrieved the pre-treatment T1
CBCT and post-intrusion T2 CBCT which were acquired three months apart. We used
OnDemand3D software to perform the superimposition, as shown in additional file movie S1.
All the bone above the teeth’s apexes was selected as stable structure because the two images
were acquired within a short period of time to each other.
Measurements of the displacements can be performed directly on the sagittal slices, providing
a quantitative evaluation of the clinical results. As shown in figure 7, we could confirm that
the main displacement of the teeth was proclination through uncontrolled tipping and slight
intrusion of teeth #11 and #21.

Patient #3, surgical orthodontic treatment (OnDemand3D software)

A 24-year-old patient was referred for open bite and skeletal Class II due to a retrognathic
mandible (figure 8a). The patient underwent a surgical orthodontic treatment. The surgery
consisted of a bi-maxillary surgery with mandibular advancement and three piece Le Fort I
osteotomy (maxillary expansion and anterior segmental osteotomy) (figure 8b).
In order to perform a 3D general superimposition, we retrieved two CT-Scan images: T1 scan
acquired one month before surgery and T2 scan acquired immediately after surgery. We used
OnDemand3D software, following the procedure described previously. We selected the
cranial base as stable structure. Results of the superimposition can be analysed directly in the
software, using transparent overlays and measurement tools in the axial slices. For example, it
is possible to evaluate the osseous displacement on sagittal views as shown in figure 9.
To visualize the overall results, we generated a 3D colour-coded mapping of the
displacements, which displays the Euclidean distances between the two images through a
colour scale (figure 10). To this aim, we segmented the T1 and T2 superimposed images and
exported the resulting 3D surface model STL files in 3D Slicer software. Then, we used the
ShapePopulationViewer extension to compute the colour-coded maps. As expected, the main
displacement was found around the chin region and the incisal edges of maxillary incisors.

Discussion
-This article focused on voxel-based 3D superimposition of craniofacial images, detailing its
implementation methods and illustrating the feasibility with 3 original clinical cases. Our
main goal was to show that 3D superimposition was efficient and could be easily performed
by sparsely-trained clinicians, especially using commercially-available software. In the near
future, this procedure should ideally become a standard approach, insofar as a clinician has
two 3D scans of a same patient, thus helping to better understand our treatment effects and
growth-related changes.
The main challenge remains the evaluation of 3D superimposition results. For now, the
easiest way to assess displacements between two images is a qualitative evaluation. It can be

5
done by using 3D models (figure 5) or by navigating in 2D slices of the superimposed images
(figures 4, 7, 9 and additional file movie S2). But these qualitative results can be hard to
interpret, remain subjective, operator dependent and cannot be strictly compared [29].

-Quantitative evaluation of 3D superimposition results is still under development and stays an

active research field. Measurement of distances and angles can be performed between
landmarks localized on the superimposed images, either on 2D slices (Figures 7,9) or on 3D
models. All the available methods offer the possibility to do so. For clinicians, measurements
of interest are the anteroposterior, vertical or transversal displacements between two
landmarks. In order to obtain these measurements, the 3D distance between the points needs
to be decomposed in the 3 planes of the space (x-, y- and z- axis components). Orientation of
the images is critical when considering these components, as a reorientation of the images
modifies their values [30]. To provide relevant descriptive results, the 3D images must be
strictly repositioned, using a common reference system, before performing measurements
[31]. To be accurate, this method requires precise localization of landmarks, which can be
time-consuming and operator-dependent [10]. In this way, computer-aided 3D landmarking
may improve accuracy and reproducibility of measurements derived from CBCT data [32].
-A current quantitative evaluation approach is based on the colour-coded map of closest
points (figure 10), [6-8]. In our knowledge, distance mapping cannot be directly performed in
commercially-available software so far. This method has the advantage to be easily readable
and automatically provides quantitative results without landmark annotation. However, these
maps do not display the distance between corresponding points but the distance between
closest points. Consequently, this method can induce distorted results in case of rotations or
important translations and should be used with caution [33].
-The aforementioned limits of colour-coded maps of closest points could be overcome by
shape correspondence analysis. The principle is to measure distances between corresponding
points by mapping the shape of the two superimposed models. This method is very promising
as it allows precise quantification of the displacements and has been shown to be effective in
mapping bone remodelling [29,33]. So far, the application of this technique is limited to the
research field. The reported method requires approximately 15 hours per patient after a
learning curve of 6 months [29]. In order to provide descriptive results in the anteroposterior,
vertical or transversal directions, repositioning of the images using a common reference
system is still needed [31].
-As with any radiographic examination, the three basic principles of radioprotection
(justification, optimization and limitation) must be strictly observed, especially for children
for whom the authors recommend referring to the DIMITRA guidelines [34]. Large field of
view (FOV) CBCT acquisitions should not be used routinely for orthodontic diagnosis and
should be restricted to complex cases of skeletal abnormality [35]. These recommendations
could evolve with the development of low dose protocols [36], on which voxel-based
superimposition technique still need to be evaluated. For instance, an in vitro study recently
showed that the accuracy of the technique can be affected by changing CBCT settings such as
kilovolt peak (kVp), milliamperage (mA) or voxel size [37].
-Each CBCT scan should be optimized individually according to the ALADAIP principle (As
Low as Diagnostically Acceptable being Indication-oriented and Patient-specific), taking into
account the diagnostic purposes as well as the surgical planning and/or 3D printing specific
applications [34]. To superimpose CBCT scans of two time-points the FOV needs to cover
the stable anatomical areas defined as reference regions, such as the cranial base for general
superimpositions. The latter may lead to a significant extension of the FOV which may
dramatically increase the radiation dose [38]. In this case, regional superimpositions should
be preferred. Overall, exposure settings should be carefully adjusted in balance with the

6
clinical aims, which may significantly reduce the radiation dose without negatively impact the
diagnosis [39].

Conclusion
If a CBCT examination is performed, the available 3D data should be exploited entirely for
the benefit of the patient. Craniofacial three-dimensional voxel-based superimpositions have
been proposed nearly 15 years ago as a convincing method to objectively monitor over time
dentofacial treatment outcomes as well as growth changes. However, this approach remains
barely used in daily clinical practice. This article aimed to make 3D superimposition
accessible to the clinician, whether through a commercially-available software or through a
combination of open source programs.

GLOSSARY
- 3D: three-Dimensional
- 2D: two-Dimensional
- DCBIA: Dental and Craniofacial Bionetwork for Image Analysis
- DICOM: Digital Imaging and Communication of Medicine
- CT: Computed Tomography
- CBCT: Cone Beam Computed Tomography
- TAD: Temporary Anchorage Device
- FOV: Field Of View
- ALADAIP: As Low As Diagnostically Acceptable being Indication-oriented
andSatient-specific
- kVp: kilovolt peak
- mA: milliamperage

Additional files
- Additional file 1: Movie S1 (.mp4): OnDemand3D software performs maxillary 3D
superimposition of two CBCT scans.
- Additional file 2: Movie S2 (.mp4): Evaluation of bone grafting results in patient #1
by navigating in axial slices of T1 and T2 superimposed 3D dataset.

Legends for tables

Table I: Available methods for voxel-based 3D superimposition of craniofacial structures

Legends for illustrations

Figure 1: Rigid registration of a cranial base 3D surface model on a target model. a: initial
position with blue model to register on grey model; b: after rotation around x axis; c: after
rotation around y axis; d: after rotation around z axis; e: after translation along x axis; f: after
translation along y axis; g: after translation along z axis.
Figure 2: Screenshot of a YouTube video tutorial freely distributed by the DCBIA to learn
how to perform 3D voxel-based superimpositions.
Figure 3: Screenshot of Supplementary material Video S1: maxillary 3D superimposition of
two CBCT scans performed with OnDemand3D software.
Figure 4: Screenshot of Supplementary material Video S2: evaluation of bone grafting results
in patient #1 by navigating in axial slices of T1 and T2 superimposed 3D dataset.

7
Figure 5: Superimposition of 3D surface models of patient #1 T1 (red) and T2 (transparent
blue) 3D datasets: a. superior view; b. left lateral view.
Figure 6: Intraoral views of patient #2. a: Before treatment; b: Four months after treatment
start.
Figure 7: Measurement of patient #2 clinical displacement of teeth #11, #21, #22 on sagittal
views, between T1 (red overlay) and T2 (white).
Figure 8: Intraoral views of patient #3. a: Before treatment; b: Two months after surgery.
Figure 9: Evaluation of patient #3 osseous displacement on a sagittal view, T2 scan (red
colour) superimposed on T1 scan (grey colour).
Figure 10: Patient #3 colour-coded map of closest points between T1 and T2 segmented 3D
dataset, after superimposition.

Declarations
Ethics approval and consent to participate: The study was performed in accordance with the Declaration of
Helsinki. Written informed consent to participate was obtained from each patient (or their parents for minor
patients).
Consent for publication: Written consent to publish all information contained in this article and any
accompanying images has been obtained from each patient (or their parents for minor patients).
Availability of data and material: The datasets analysed during the current study are not publicly available due
to privacy reasons, but are available from the corresponding author on reasonable request.
Disclosure of interest: The authors declare that they have no competing interest.
Funding: No specific funding was used for this study. Cybermed Inc. (Seoul, Korea) provided a free trial
licence of OnDemand3D software.

REFERENCES
[1] Gu Y, McNamara JA. Cephalometric Superimpositions: A Comparison of Anatomical
and Metallic Implant Methods. Angle Orthod 2008;78:967–76.
https://doi.org/10.2319/070107-301.1.
[2] American Board of Orthodontics n.d. https://www.americanboardortho.com/ (accessed
October 25, 2019).
[3] EBO n.d. https://www.eoseurope.org/ebo/ebo (accessed October 25, 2019).
[4] Gribel BF, Gribel MN, Frazäo DC, McNamara JA, Manzi FR. Accuracy and
reliability of craniometric measurements on lateral cephalometry and 3D measurements on
CBCT scans. Angle Orthod 2011;81:26–35. https://doi.org/10.2319/032210-166.1.
[5] Oliveira FPM, Tavares JMRS. Medical image registration: a review. Comput Methods
Biomech Biomed Engin 2014;17:73–93. https://doi.org/10.1080/10255842.2012.670855.
[6] Cevidanes, L, Bailey L, Tucker G, Styner M, Mol A, Phillips C, et al. Superimposition
of 3D cone-beam CT models of orthognathic surgery patients. Dentomaxillofacial Radiol
2005;34:369–75. https://doi.org/10.1259/dmfr/17102411.
[7] Cevidanes LHC, Heymann G, Cornelis MA, DeClerck HJ, Tulloch JFC.
Superimposition of 3-dimensional cone-beam computed tomography models of growing
patients. Am J Orthod Dentofacial Orthop 2009;136:94–9.
https://doi.org/10.1016/j.ajodo.2009.01.018.
[8] Yatabe M, Prieto JC, Styner M, Zhu H, Ruellas AC, Paniagua B, et al. 3D
superimposition of craniofacial imaging—The utility of multicentre collaborations. Orthod
Craniofac Res 2019;22:213–20. https://doi.org/10.1111/ocr.12281.
[9] Ponce-Garcia C, Lagravere-Vich M, Cevidanes LHS, de Olivera Ruellas AC, Carey J,
Flores-Mir C. Reliability of three-dimensional anterior cranial base superimposition methods
for assessment of overall hard tissue changes: A systematic review. Angle Orthod
2018;88:233–45. https://doi.org/10.2319/071217-468.1.
[10] Lagravère MO, Low C, Flores-Mir C, Chung R, Carey JP, Heo G, et al. Intraexaminer

8
and interexaminer reliabilities of landmark identification on digitized lateral cephalograms
and formatted 3-dimensional cone-beam computerized tomography images. Am J Orthod
Dentofacial Orthop 2010;137:598–604. https://doi.org/10.1016/j.ajodo.2008.07.018.
[11] Ponce-Garcia C, Ruellas AC de O, Cevidanes LHS, Flores-Mir C, Carey JP,
Lagravere-Vich M. Measurement error and reliability of three available 3D superimposition
methods in growing patients. Head Face Med 2020;16:1. https://doi.org/10.1186/s13005-020-
0215-7.
[12] Almukhtar A, Ju X, Khambay B, McDonald J, Ayoub A. Comparison of the Accuracy
of Voxel Based Registration and Surface Based Registration for 3D Assessment of Surgical
Change following Orthognathic Surgery. PLoS One 2014;9:e93402.
https://doi.org/10.1371/journal.pone.0093402.
[13] Maes F, Collignon A, Vandermeulen D, Marchal G, Suetens P. Multimodality image
registration by maximization of mutual information. IEEE Trans Med Imaging 1997;16:187–
98. https://doi.org/10.1109/42.563664.
[14] Ruellas ACO, Huanca Ghislanzoni LT, Gomes MR, Danesi C, Lione R, Nguyen T, et
al. Comparison and reproducibility of 2 regions of reference for maxillary regional
registration with cone-beam computed tomography. Am J Orthod Dentofacial Orthop
2016;149:533–42. https://doi.org/10.1016/j.ajodo.2015.09.026.
[15] Ruellas AC de O, Yatabe MS, Souki BQ, Benavides E, Nguyen T, Luiz RR, et al. 3D
Mandibular Superimposition: Comparison of Regions of Reference for Voxel-Based
Registration. PLOS One 2016;11:e0157625. https://doi.org/10.1371/journal.pone.0157625.
[16] Nguyen T, Cevidanes L, Franchi L, Ruellas A, Jackson T. Three-dimensional
mandibular regional superimposition in growing patients. Am J Orthod Dentofacial Orthop
2018;153:747–54. https://doi.org/10.1016/j.ajodo.2017.07.026.
[17] Schilling J, Gomes LCR, Benavides E, Nguyen T, Paniagua B, Styner M, et al.
Regional 3D superimposition to assess temporomandibular joint condylar morphology.
Dentomaxillofacial Radiol 2014;43:20130273. https://doi.org/10.1259/dmfr.20130273.
[18] DCBIA Video Tutorials n.d. https://sites.google.com/a/umich.edu/dentistry-image-
computing/e-learning/Videos-Tutorials (accessed October 25, 2019).
[19] ITK-SNAP n.d. http://www.itksnap.org/pmwiki/pmwiki.php (accessed October 25,
2019).
[20] 3D Slicer n.d. https://www.slicer.org/ (accessed October 25, 2019).
[21] Bazina M, Cevidanes L, Ruellas A, Valiathan M, Quereshy F, Syed A, et al. Precision
and reliability of Dolphin 3-dimensional voxel-based superimposition. Am J Orthod
Dentofacial Orthop 2018;153:599–606. https://doi.org/10.1016/j.ajodo.2017.07.025.
[22] Nada RM, Maal TJJ, Breuning KH, Bergé SJ, Mostafa YA, Kuijpers-Jagtman AM.
Accuracy and Reproducibility of Voxel Based Superimposition of Cone Beam Computed
Tomography Models on the Anterior Cranial Base and the Zygomatic Arches. PLoS ONE
2011;6:e16520. https://doi.org/10.1371/journal.pone.0016520.
[23] Häner ST, Kanavakis G, Matthey F, Gkantidis N. Voxel‐based superimposition of
serial craniofacial CBCTs: Reliability, reproducibility and segmentation effect on hard‐tissue
outcomes. Orthod Craniofac Res 2019:ocr.12347. https://doi.org/10.1111/ocr.12347.
[24] Weissheimer A, Menezes LM, Koerich L, Pham J, Cevidanes LHS. Fast three-
dimensional superimposition of cone beam computed tomography for orthopaedics and
orthognathic surgery evaluation. Int J Oral Maxillofac Surg 2015;44:1188–96.
https://doi.org/10.1016/j.ijom.2015.04.001.
[25] Lee J-H, Kim M-J, Kim S-M, Kwon O-H, Kim Y-K. The 3D CT superimposition
method using image fusion based on the maximum mutual information algorithm for the
assessment of oral and maxillofacial surgery treatment results. Oral Surg Oral Med Oral
Pathol Oral Radiol 2012;114:167–74. https://doi.org/10.1016/j.tripleo.2011.06.003.

9
[26] Koerich L, Weissheimer A, de Menezes LM, Lindauer SJ. Rapid 3D mandibular
superimposition for growing patients. Angle Orthod 2017;87:473–9.
https://doi.org/10.2319/072316-574.1.
[27] Koerich L, Burns D, Weissheimer A, Claus JDP. Three-dimensional maxillary and
mandibular regional superimposition using cone beam computed tomography: a validation
study. Int J Oral Maxillofac Surg 2016;45:662–9. https://doi.org/10.1016/j.ijom.2015.12.006.
[28] Choi J-H, Mah J. A new method for superimposition of CBCT volumes. J Clin Orthod
JCO 2010;44:303–12.
[29] Yatabe M, Garib D, Faco R, de Clerck H, Souki B, Janson G, et al. Mandibular and
glenoid fossa changes after bone-anchored maxillary protraction therapy in patients with
UCLP: A 3-D preliminary assessment. Angle Orthod 2016;87:423–31.
https://doi.org/10.2319/052516-419.1.
[30] Ruellas AC de O, Tonello C, Gomes LR, Yatabe MS, Macron L, Lopinto J, et al.
Common 3-dimensional coordinate system for assessment of directional changes. Am J
Orthod Dentofacial Orthop 2016;149:645–56. https://doi.org/10.1016/j.ajodo.2015.10.021.
[31] Shahen S, Lagravère MO, Carrino G, Fahim F, Abdelsalam R, Flores-Mir C, et al.
United Reference Method for three-dimensional treatment evaluation. Prog Orthod
2018;19:47. https://doi.org/10.1186/s40510-018-0242-0.
[32] Dot G, Rafflenbeul F, Arbotto M, Gajny L, Rouch P, Schouman T. Accuracy and
reliability of automatic three-dimensional cephalometric landmarking. Int J Oral Maxillofac
Surg 2020:S0901502720300837. https://doi.org/10.1016/j.ijom.2020.02.015.
[33] Nguyen T, Cevidanes L, Paniagua B, Zhu H, Koerich L, De Clerck H. Use of shape
correspondence analysis to quantify skeletal changes associated with bone-anchored Class III
correction. Angle Orthod 2014;84:329–36. https://doi.org/10.2319/041513-288.1.
[34] Oenning AC, Jacobs R, Pauwels R, Stratis A, Hedesiu M, Salmon B, et al. Cone-beam
CT in paediatric dentistry: DIMITRA project position statement. Pediatr Radiol 2018;48:308–
16. https://doi.org/10.1007/s00247-017-4012-9.
[35] SEDENTEXCT project. Cone Beam CT for dental and maxillofacial radiology
(Evidence based guidelines). European Commission; 2012.
[36] Ludlow JB, Timothy R, Walker C, Hunter R, Benavides E, Samuelson DB, et al.
Effective dose of dental CBCT—a meta analysis of published data and additional data for
nine CBCT units. Dentomaxillofacial Radiol 2015;44:20140197.
https://doi.org/10.1259/dmfr.20140197.
[37] Eliliwi M, Bazina M, Palomo JM. kVp, mA, and voxel size effect on 3D voxel-based
superimposition. Angle Orthod 2019:012719-52.1. https://doi.org/10.2319/012719-52.1.
[38] Marcu M, Hedesiu M, Salmon B, Pauwels R, Stratis A, Oenning ACC, et al.
Estimation of the radiation dose for pediatric CBCT indications: a prospective study on
ProMax3D. Int J Paediatr Dent 2018;28:300–9. https://doi.org/10.1111/ipd.12355.
[39] Oenning AC, Pauwels R, Stratis A, De Faria Vasconcelos K, Tijskens E, De Grauwe
A, et al. Halve the dose while maintaining image quality in paediatric Cone Beam CT. Sci
Rep 2019;9:5521. https://doi.org/10.1038/s41598-019-41949-w.

10
 DCBIA open-source method Commercially-available software methods

1- ITK-Snap (semi-automatic
segmentation of the region of
Dolphin InVivo
Software reference) OnDemand3D Maxilim
Imaging Dental
2- 3D Slicer (registration,
evaluation of the results)
Adult and Adult and
General Adult and growing patients growing growing Adult
n/a
superimposition [6,7] patients patients patients[22]
Validation [24,25] [21,23]
studies Adult and
Regional Adult and growing patients growing
n/a n/a n/a
superimpositions [14–17] patients
[26,27]

General Non-growing patients: entire cranial base [6]

Region of superimposition Growing patients: anterior cranial base (growth completed by age 7)[7,9]

reference
Regional Maxilla, growing patients: key ridge region [14]
superimpositions Mandible, growing patients: chin and symphysis regions [16]

1- Approximate manual
registration of the images 1- Approximate manual registration of the images
2- Segmentation of the region 2- Selection of the region of reference
Main steps
of reference 3- Algorithm launch
3- Algorithm launch 4- Evaluation of the results
4- Evaluation of the results

Several hours, series of 20

Operator training Less than one hour [24]
video tutorials
±3 hours for an experienced
Duration Few minutes [21,24]
operator [21]
Relies only on freely-
accessible, easily-modifiable
User-friendly [24]
Main advantages and open-source software[17]
Do not require extensive training of the operators
Color-coded maps of the
results available

Requires some computing Software license required

Main drawbacks
skills, training and time [8] No color-coded maps of the results available

Main applications Research Research and clinical purposes