sensors

Systematic Review
The Use and Performance of Artificial Intelligence in
Prosthodontics: A Systematic Review
Selina A. Bernauer, Nicola U. Zitzmann and Tim Joda *

UZB University Center for Dental Medicine Basel, Department of Reconstructive Dentistry, University of Basel,
4058 Basel, Switzerland; [email protected] (S.A.B.); [email protected] (N.U.Z.)
* Correspondence: [email protected]

Abstract: (1) Background: The rapid pace of digital development in everyday life is also reflected
in dentistry, including the emergence of the first systems based on artificial intelligence (AI). This
systematic review focused on the recent scientific literature and provides an overview of the ap-
plication of AI in the dental discipline of prosthodontics. (2) Method: According to a modified
PICO-strategy, an electronic (MEDLINE, EMBASE, CENTRAL) and manual search up to 30 June
2021 was carried out for the literature published in the last five years reporting the use of AI in the
field of prosthodontics. (3) Results: 560 titles were screened, of which 30 abstracts and 16 full texts
were selected for further review. Seven studies met the inclusion criteria and were analyzed. Most of
the identified studies reported the training and application of an AI system (n = 6) or explored the
function of an intrinsic AI system in a CAD software (n = 1). (4) Conclusions: While the number of
included studies reporting the use of AI was relatively low, the summary of the obtained findings
by the included studies represents the latest AI developments in prosthodontics demonstrating its
application for automated diagnostics, as a predictive measure, and as a classification or identification



tool. In the future, AI technologies will likely be used for collecting, processing, and organizing


patient-related datasets to provide patient-centered, individualized dental treatment.
Citation: Bernauer, S.A.;
Zitzmann, N.U.; Joda, T. The Use and
Keywords: artificial intelligence; machine learning; deep learning; neural networks; reconstructive
Performance of Artificial Intelligence
dentistry; prosthetic treatment
in Prosthodontics: A Systematic
Review. Sensors 2021, 21, 6628.
https://doi.org/10.3390/s21196628

Academic Editor: Yu-Dong Zhang 1. Introduction

“Digital transformations”, “digitized workflows”, “technical developments”: these
Received: 7 September 2021 terms describe some of the game changers of the 21st century, both in social life as well
Accepted: 1 October 2021 as dental medicine [1]. The use of mobile devices, tablets, and smartphones, and the
Published: 5 October 2021 easy access to technology and the World Wide Web, have changed the cultural habits of
our society in general [1]. It is not surprising that more advanced technologies, such as
Publisher’s Note: MDPI stays neutral artificial intelligence (AI), are also finding increasing application in daily life. Generally
with regard to jurisdictional claims in speaking, AI imitates the cognitive processes of human intelligence by using machines and
published maps and institutional affil- software-type algorithms to manage complex tasks [2,3].
iations.
AI applications are commonplace in digital everyday life, for example in the form
of virtual assistants such as “Siri” or “Alexa” [1,4], and AI applications are implemented
in various engineering fields [5–7]. In medicine, AI algorithms are ubiquitously used for
image processing via feature extraction of specific images and target conducting [8]. For
Copyright: © 2021 by the authors. example, AI algorithms can be used for analyzation of chest X-ray and lung CT image
Licensee MDPI, Basel, Switzerland. samples images of COVID-19 patients and thus decrease the picture diagnostic time of
This article is an open access article the radiologist and accelerate clinical decisions [9]. Furthermore, technological advances
distributed under the terms and are increasingly noticeable in dentistry, but are still restrained compared to medicine.
conditions of the Creative Commons
Digital dental processes are continuously being standardized and are becoming part of
Attribution (CC BY) license (https://
routine treatment protocols [10–12]. In particular, computer-aided design/computer-aided
creativecommons.org/licenses/by/
manufacturing (CAD/CAM) procedures have found their way into everyday clinical and
4.0/).

Sensors 2021, 21, 6628. https://doi.org/10.3390/s21196628 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 6628 2 of 11

laboratory practice [10,13]. Digitization in dentistry continues to develop and new trends
are emerging, including the application of AI.
AI is a widely used term that warrants a precise definition (Figure 1); in general,
algorithms that simulate the processes of human intelligence to resolve problems are
summarized under the term AI [4,14]. Intelligence is defined as: “the ability of a system
to act appropriately in an uncertain environment, where appropriate action is that which
increases the probability of success, and success is the achievement of behavioral subgoals
that support the system’s ultimate goal” [15]. Strong and weak AI can be differentiated.
Strong AI describes a system “that was operated in the same way as human intelligence
through non-natural, artificial hardware, and software construction” [4]. Weak AI, on the
other hand, does not aim to mimic human intelligence in its entirety, but rather is a system,
in which the advantages of medical and logical algorithms can be used by humans. Weak
AI recognizes that computer-integrated technology is different from human intellectual
performances [4].

Figure 1. Flowchart of the electronic and manual search strategy.

Inspiration for the formation of artificial networks can be found in the biology of the
human brain. “An artificial neuron is a mathematical function conceived as a coarse model
of a biological neuron. The principle is to simulate the transfer of information through
a neuron: weighted nodes receive the inputs (representing the synapses), sum them to
produce an activation (representing the axon), and pass this activation to a nonlinear
function called activation or transfer function, in order to generate the output signal.
Each neuron acts as an elementary processing unit. The output signal of one unit will
feed the other units, organized in layers, and so on, forming an artificial neural network
(ANN)” [16]. There are typically three phases in the development and application of an
ANN: (a) the training phase, where the ANN learns; (b) the validation phase, where the
ANN’s results and reference values are reconciled; and (c) the inference or application
phase, where the ANN is used in real cases [14].
Machine learning (ML) is part of weak AI; by recognizing patterns, the computer
system is able to learn and make predictions [4,17]. After being trained on existing datasets,
the ML system is able to provide accurate predictions on new data. Often the training
datasets need to be simplified, so that the algorithm can recognize the desired pattern.
Depending on how carefully these interventions are made, the performance of ML systems
varies [16]. Current examples of ML systems include translators such as DeepL, diagnosis
Sensors 2021, 21, 6628 3 of 11

tools, and speech recognition frameworks [18]. Deep learning (DL) is part of ML that
describes “a set of computational models composed of multiple layers of data processing,
which make it possible to learn by representing these data through several levels of abstrac-
tion” [16]. Compared to ML, deep neural networks (DNNs) are able to independently learn
and hierarchize the training dataset but require, therefore, larger amounts of data. It comes
with the advantage that new feature extractors are not needed for every problem [16].
DL works via a backpropagation algorithm to train neural networks [16]. Convolutional
neural networks (CNN) are a special part of DL, used especially for image processing and
analyses of radiological datasets [16].
AI technologies are well suited for repetitive work using large datasets. The greater
the workload, the more precise the performance of AI becomes [1]. Since dentistry provides
many areas for assisting work and automation of simple routine tasks, AI could find
applications here and support dentists to improve quality of work and accuracy [18].
The domain of AI in (dental) medicine is image data processing in radiological work-
flows, especially for detection of caries and periapical endodontic lesions [16], as well as
automatic identification and classification of oral implant systems [19]. In prosthodontics,
however, AI technology is still scarce due to complex diagnostics and treatment requiring
individual protocols.
Therefore, the aim of this systematic review was to analyze the recent scientific
literature on the diagnostic performance and the clinical applications of AI in applied
prosthodontics. The remainder of this systematic review was structured as follows:
(a) introduction of the search strategy and related inclusion criteria for data extraction;
(b) summary of the obtained findings of the identified and included studies; (c) critical
discussion of the results; (d) conclusions.

2. Materials and Methods

This systematic review was conducted in accordance with the guidelines of PRISMA
(Preferred Reporting Items of Systematic Reviews and Meta-Analyses) [20].

2.1. Search Strategy

Based on the PICO criteria, a search strategy was developed and an electronic search
was conducted. The PICO question was formulated as follows: What are the current clinical
applications and diagnostic performance of AI in the field of prosthodontics?
A systematic electronic search of PubMed MEDLINE, EMBASE, and CENTRAL was
performed for English-language publications; in addition, Google Scholar was reviewed.
Search syntax was categorized into population, intervention, comparison, and outcome
(PICO). Each category was assembled from a combination of Medical Subject Headings
(MeSH Terms) as well as free-text words in simple or multiple conjunctions: ((Prosthodon-
tics [Mesh]) OR (prosthetic treatment) OR (reconstructive therapy)) AND ((Artificial Intelli-
gence [Mesh]) OR (Machine Learning OR Deep Learning OR Neural Networks [Mesh]))
(Table 1).
Additional manual searches of the bibliographies of all full-text articles and related
reviews selected from the electronic search were also performed. Furthermore, manual
searching was conducted in the following journals: Journal of Prosthodontic Research,
Journal of Prosthetic Dentistry, Clinical Oral Implants Research, International Journal of
Oral Maxillofacial Implants, Clinical Implant Dentistry and Related Research, Implant
Dentistry, and Journal of Implantology.
Sensors 2021, 21, 6628 4 of 11

2.2. Inclusion and Exclusion Criteria

Inclusion criteria for the studies were defined as follows:
• Studies at all levels of evidence, except expert opinion;
• Articles published in English;
• Articles published in the last 5 years (up to 30 June 2021).
Exclusion criteria for the studies were defined as follows:
• Review articles, letter to editors and case reports involving less than 5 cases;
• Animal studies;
• Full-text not available/accessible.

Table 1. Search strategy according to the PICO criteria.

Focused Question What Are the Current Clinical Applications and Diagnostic Performance of Artificial Intelligence (AI)
(PICO) in Prosthodontics?
Patients with indication for prosthetic reconstructions
Search Strategy Population
#1 ((Prosthodontics [Mesh]) OR (prosthetic treatment) OR (reconstructive therapy))
Diagnostic model based on applied AI algorithms
Intervention or
#2 ((Artificial Intelligence [Mesh]) OR (Machine Learning OR Deep Learning OR
exposure
Neural Networks [Mesh]))
Comparison N.A.
Outcome Clinical applications or diagnostic performance of the proposed AI model
Search #1 AND #2
combination Limitations: Articles published in the last 5 years (up to 30 June 2021); English
Database search Electronic PubMed Medline, Embase, Central, manual search
Journal of Prosthodontic Research, Journal of Prosthetic Dentistry, Clinical Oral
Journals Implants Research, International Journal of Oral Maxillofacial Implants, Clinical
Implant Dentistry and Related Research, Implant Dentistry, Journal of Implantology
Studies at all levels of evidence, except expert opinion;
Selection criteria Inclusion criteria
Articles published in English;Articles published in the last 5 years.
Review articles, letter to editors and case reports/case series involving less than
5 cases;
Exclusion criteria Animal studies;
Multiple publications on the same patient population;
Full text not available/accessible.

2.3. Data Extraction

Two reviewers (S.B. and T.J.) independently screened the titles and abstracts that
were identified in the searches according to the defined inclusion and exclusion criteria.
If sufficient information could not be extracted from titles and abstracts, the full text was
consulted. The full texts of potentially relevant articles were obtained and reviewed in
detail by both reviewers, from which the final list of articles was selected for further
analysis. Disagreements were resolved by discussion.
The following information was collected from selected articles:
• Author(s), year of publication, country, study design;
• Total number of patients/datasets;
• Training/validation datasets;
• Test datasets;
• Aim of the study;
• AI application; and
• Outcome.
The information extracted from the articles was tabulated. Assessment of risk of
bias was carried out for each included study, using the Newcastle–Ottawa Assessment
Scale (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp (Accessed on 7
September 2021)) to assess the quality of the studies (Table 2).
Sensors 2021, 21, 6628 5 of 11

Table 2. Presentation of risk of bias evaluation for included studies.

Selection Comparability Outcome

(Max. 4 Stars) (Max. 2 Stars) (Max. 4 Stars)
Lee, J.H. et al. (2020) ** − *
Lerner, H. et al. (2020) *** − *
Yamaguchi, S. et al. (2019) *** * *
Lee, J.H. et al. (2018a) ** − *
Lee, J.H. et al. (2018b) ** − *
Raith, S. et al. (2017) ** − *
Wei, J. et al. (2016) * − *

3. Results
3.1. Included Studies
The systematic search was finished on 30 June 2021. A total of 560 article titles were
screened and 30 abstracts were considered for further analysis. Subsequently, the full texts
of 16 articles were analyzed to determine if they met the inclusion criteria. After further
examination, 11 articles were excluded due to the following reasons:
• Not a study in the field of AI application in prosthodontics (n = 6);
• Not a clinical study (n = 3);
• Full text not available (n = 1);
• Missing information on AI technology (n = 1).
Moreover, two other articles meeting the inclusion criteria were added from manual
searching; subsequently, a total of seven full-text articles were included for data extraction
(Figure 1).

3.2. Descriptive Analysis

The seven included full-text articles came from diverse areas of the field of prosthodon-
tics and demonstrated a wide range of the use of AI. Six studies investigated the training
and application of different AI systems [19,22–26] and one study explored the function
of an intrinsic AI system of a specific CAD software for designing prosthetic reconstruc-
tions [21]. Direct comparisons of the identified and included studies were not feasible
because of the heterogeneity of the specific aims, defined outcomes, and topics within
the field of reconstructive dentistry. Therefore, the analysis of the information reported
by the included studies follows a descriptive approach. The detailed data extraction is
summarized in Table 3.
All included studies were designed as non-randomized, retrospective cohort studies.
Three of the publications were from one research group with a special focus on radiographic
image analysis [19,23,24]. The number of datasets used for AI training in the studies varied
considerably (from 43 to 10770). Five of the studies included investigated CNN models,
one study applied an ANN model, and another study used an intrinsic AI and algorithms
of a commercially available laboratory CAD software.
Sensors 2021, 21, 6628 6 of 11

Table 3. Characteristics and outcomes of the studies included.

Training/
First Author Test
Study Design n Datasets Validation Aim of the Study AI Application Outcome
(Year) Country Datasets
Datasets
“The aim of the current study “Deep CNN architecture is
was to evaluate the efficacy of useful for the identification
10,770
Lee (2020) [19] Retrospective 6462 (60%) deep CNN algorithm for the CNN and classification of dental
radiographic 2154 (20%)
Korea cohort study 2154 (20%) identification and (GoogLeNetInception v3) implant systems using
images
classification of dental panoramic and periapical
implant systems”. radiographic images”.
“Purpose of this retrospective “Using intrinsic AI, the
clinical study is to present a software was able to
protocol for the use of AI to Intrinsic AI and algorithms automatically trace the margin
Lerner (2020) [21] Retrospective 106 fabricate implant-supported of the CAD software line of the implant abutment,
n.r. n.r.
Germany cohort study restorations monolithic zirconia crowns (Valletta® , Exocad, though subgingival”.
cemented on customized Darmstadt, Germany) In 96.2% of the restorations,
hybrid abutments, via a full the marginal adaption was
digital workflow”. very accurate.
“The aim of this study was to
assess the validity of deep
High performance of AI in
learning with a CNN method CNN; implemented with the
predicting the debonding
to predict the debonding Keras library (version 2.2.4)
Yamaguchi (2019) Retrospective probability of 2160 test
8640 6480 2160 probability of CAD/CAM on top of TensorFlow (GPU
[22] Japan cohort study 2D-images of CAD/CAM
composite resins restorations version 1.12.2) in Python
crowns with a current
from 2D images captured from (version 3.7.2)
prediction accuracy of 98.5%”.
3D STL models of a die
scanned by a 3D oral scanner”.
High diagnostic accuracies of
“The aim of the current study premolar (89%), molar (88%)
3000 was to evaluate the efficacy of and both premolar and molar
Lee (2018a) [23] Retrospective periapical deep CNN algorithms for CNN (82%) models were achieved.
2400 (80%) 600 (20%)
Korea cohort study radiographic detection and diagnosis of (GoogLeNetInception v3) “Deep CNN algorithms are
images dental caries on expected to be among the most
periapical radiographs”. effective and efficient methods
for diagnosing dental caries”.
Sensors 2021, 21, 6628 7 of 11

Table 3. Cont.

Training/
First Author Test
Study Design n Datasets Validation Aim of the Study AI Application Outcome
(Year) Country Datasets
Datasets
“The aim of the current study
“With the deep learning
was to develop a
algorithm, the diagnostic
computer-assisted detection
accuracy for periodontally
1740 system based on a deep CNN
compromised teeth was 81.0%
Lee (2018b) [24] Retrospective periapical algorithm and to evaluate the CNN; based on a Keras
1044 348 for premolars and 76.7% for
Korea cohort study radiographic potential usefulness and framework in Python
molars. [ . . . ] The deep CNN
images accuracy of this system for the
algorithm was useful for
diagnosis and prediction of
assessing the diagnosis and
periodontally
predictability of PCT”.
compromised teeth”.
High performance with correct
“[The] hypothesis is that tooth classifications were shown;
classification algorithms based ANN; principal algorithm “cusps are detected
on ANNs are capable of based on blob detection with automatically and thus
Raith (2016) [25] Retrospective classifying teeth with sufficient a Difference of Gaussians completely reproducible,
129 datasets n.r. n.r.
Germany cohort study accuracy for potential use in (DoG) approach, which is advantageous when
clinical practice in order to implemented in Python standardized treatment
improve digital workflow in programming language concepts need to be
dental prosthetics”. established, paving the way
for evidence-based dentistry”.
“[The aim of this study was] to
explore the feasibility of a
CNN; back-propagation “The novel computer color
novel computer
neural network (BPNN) is a matching system produced
color-matching system based
Wei (2016) [26] Retrospective multilayer feed-forward greater accuracy in color
43 datasets 39 4 on the improved
China cohort study neural network trained by reproduction within the given
back-propagation neural
the error back- color space than the traditional
network model by comparing
propagation algorithms visual approach”.
it with the traditional
visual method”.
AI: artificial intelligence; ANN: artificial neural network; CNN: convolutional neural networks; n: number; n.r.: not reported.
Sensors 2021, 21, 6628 8 of 11

4. Discussion
This systematic review focused on applied AI technologies in prosthodontics, and it
was demonstrated that AI was used for automated diagnostics, as a predictive measure,
and as a classification or identification tool. The findings indicated that, in the wider field
of prosthodontics, AI has been applied to CAD/CAM systems, implant prosthetics, tooth
preservation, and orofacial anatomy. Since digital technologies are developing rapidly and
the observed turnover rate of obsolete software is about 1.5–2 years, the current search was
limited to the last 5 years. The aim of this systematic review was not to provide a historical
overview of AI technology in dentistry in general. Rather, the manuscript focused on
prosthetic AI applications. Furthermore, the systematic search focused on clinical trials and
case series with at least 5 patients to increase the scientific level (and to avoid including
technical reports).
The overall number of eligible studies investigating AI applications in prosthodontics
was relatively low (n = 7). Although automated detection of caries has been investigated in
AI dental imaging diagnostics a few years ago, the use of AI technology in prosthodontics
is (still) rare. Prosthodontics itself is a diverse and complex area of dental medicine and
one that may benefit from the routine application of AI technologies. Long-term success
relies on good prognosis of abutment teeth (and/or implants) in all facets of periodontal,
endodontic, operative and reconstructive principles, and include patient-specific factors,
such as load situation, personal and medical issues, and supportive care [27]. Successful
prosthetic reconstructions require the use of a synoptic treatment concept with sufficient
backward planning and a clean practical implementation, including dental laboratory
workflows. Furthermore, the various different patient-related factors must be taken into
account; many intraoral situations are not directly comparable, with the exception of
edentulous patients. One factor is that the number of teeth to be replaced varies extremely
from 1 to 28. Mathematically, the number of possible situations, starting from 28 teeth,
would be the factorial of 27, i.e., 27×, 26×, 25×, 24×, etc. The number of options, combined
with the factors mentioned above, makes prosthetic therapy a demanding and complex
specialty, which requires a certain knowledge in adjacent disciplines. AI technologies
are particularly suited to dealing with complex situations with multiple possible factors;
therefore, the application of AI to prosthetic workflows is of high interest.
This systematic review revealed that AI systems are currently mainly limited to test
versions of automated diagnostics, especially in dental imaging and radiology [3], and
classification tools such as for periodontally compromised teeth, dental cusps, or caries.
Encouragingly, the identified studies all reported high performance of the different AI
systems investigated (Table 3), i.e., including high diagnostic accuracy for dental caries [23],
and very good prediction accuracy for debonding of CAD/CAM crowns [22].
AI is increasingly being applied to dentistry including in diverse areas of prosthetic
research for efficient data processing. The first application of AI within dentistry was in the
classification of diagnostic images and processing of data from surface scanning techniques,
because the digitally coded images could easily be transferred into AI systems [3,17].
The application of AI in diagnostics continues to be developed. AI technologies in den-
tistry have the power to become central in the triad of patient data management, health
care application, and services, and can facilitate future developments in patient-centered
individualized treatment [3].
Beyond prosthodontics, AI was previously linked to other dental disciplines [2]. In
tooth preservation, radiologically driven AI analyses can help detect root fractures and
identify periapical pathologies [28] or classify root morphologies [29]. In periodontology,
disease progression can be evaluated while clinical and radiological periodontal parameters
are automatically determined following AI technology. In oral surgery, AI can be used to
screen radiological images for pathological changes, such as cysts and bony tumors [2].
Furthermore, there are possible applications in implantology. AI-based treatment planning
in CAD/CAM implant dentistry could be of great interest in order to simplify virtual
Sensors 2021, 21, 6628 9 of 11

3D treatment planning, and, consecutively, robotic insertion of dental implants using AI

applications [30].
AI will certainly play a significant role in dentistry in the future and the development
of these technologies is awaited with excitement. AI has a disruptive potential to renew
processes in all fields of dentistry; but, due to the complexity of prosthetic treatment
concepts, the adoption of AI technology in prosthodontics is still rather hesitant. AI
systems are particularly beneficial for processing and analyzing large amounts of data
to classify outcomes, and for processing repetitive workflows. AI algorithms will likely
provide support in evidence-based dental decision making, particularly for less experienced
practitioners, and facilitate the analysis of individual patient cases. More homogeneous
treatment protocols could be ensured that still allow for individualized and personalized
treatment. Chen et al. developed a prototype decision model to assist (unexperienced)
dentists in choosing appropriate removable prosthetic options [31]. Such supportive
tools in therapy planning of (complex) patient cases in prosthodontics enable the further
development of tele-dentistry.
The combination of AI technologies in the field of prosthodontics could lead to a wide
variety of novel options, such as AI systems for generating occlusal surface design for
crowns accounting for existing intraoral wear facets, as automatic set-up designers for
complete dentures, for determining the emergence profile in implantology, or in automatic
framework designs for removable partial dentures [32]. Finally, as an educational tool, AI
already provides the opportunity to support less experienced undergraduate students in
their professional development [33].
In principle, there are no limitations to using the power of AI in prosthodontics. In
many cases, however, it is (unfortunately) financial factors that trigger the development
and standard implementation of new AI technologies in the most economically profitable
areas—and tend to prevent interesting dental applications because a market appears too
small for the dental industry. Nevertheless, future developments and research on AI can
be eagerly awaited.
Although this seems to be very promising, the obstacles of AI systems should not
be ignored. Based on this systematic literature search, the identified trials must be seen
as pilot studies and experimental, and the technologies used in these studies are not yet
suitable for everyday clinical routine in the focus of prosthodontics. At the moment, AI is
mainly used in undergraduate dental education and for academic research. Improvements
in the technology and user interfaces is needed before the techniques can be implemented
into routine dental practice. Some intermediate steps still need to be developed to make AI
attractive and helpful for its routine implementation. Beyond the technical possibilities,
the cost–benefit ratio and ethical aspects of the use of patient-specific data also need to be
critically examined in future research on AI technology in (reconstructive) dentistry [3].

5. Conclusions
While the number of studies reporting the use of AI in prosthodontics identified in
this systematic review was relatively low (inclusion of seven out of sixteen full-texts), these
studies reflect an honest overview of the latest developments in AI focusing on diagnostics,
predictive measures, classification, and identification tools. Prosthodontics is the dental
discipline with the greatest variance in terms of diagnostics and, in particular, treatment
options. Therefore, it does not seem surprising that the results of this systematic review are
comparably diverse and heterogeneous.
In addition, dentistry is (still) lagging behind medicine in terms of routine use of AI
technology. Nevertheless, AI applications in prosthodontics have the potential to open
up a wide range of opportunities for clinicians and patients, and could be used as a
supplementary future basic tool for collecting, processing, and organizing patient-related
datasets to provide patient-centered, individualized, and personalized treatment. Possible
applications of AI algorithms are very diverse in prosthodontics and one can eagerly await
further research foci and developments.
Sensors 2021, 21, 6628 10 of 11

The step-by-step implementation of digital applications (in dentistry) is linked to

the necessary technical development and is dependent on the flexibility of the users in
everyday life. The willingness to learn new treatment protocols and to trust computerized
applications has been proven as a strong negative driver for dentists [34]. In addition,
new technologies require continuous investment of the dental community. As long as
the research results are not demonstrating superiority, the routine implementation of AI
applications in prosthodontics will be delayed. Here, the university and dental schools
need to foster AI technology in research and education.
A critical topic and crucial factor for the successful rehabilitation of complex pros-
thetic cases is the correct clinical definition of vertical and horizontal maxillo-mandibular
relationships. AI technology could be used for automatic registering of jaw relationships
based on radiological landmarks in cone-beam computed tomography. In this context, the
configuration of virtual dental articulators could be synchronized with the radiological
situation to simulate individual patient movements for treatment simulation and final
fabrication of prosthetic reconstructions.
The use of AI technologies in prosthodontics is conceivable in many ways (and ab-
solutely desirable) from the clinicians’ point of view, but not least the demand and the
economic efficiency will decide whether the MedTech industry will push AI in prosthodon-
tics as fast and to what extent.

Author Contributions: Conceptualization, S.A.B. and T.J.; methodology, S.A.B. and T.J.; formal
analysis, S.A.B. and T.J.; investigation, S.A.B. and T.J.; data curation, S.A.B. and T.J.; writing—
original draft preparation, S.A.B.; writing—review and editing, T.J. and N.U.Z.; visualization, S.A.B.;
supervision, T.J. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The data are not publicly available due to privacy restrictions.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Joda, T.; Bornstein, M.M.; Jung, R.E.; Ferrari, M.; Waltimo, T.; Zitzmann, N.U. Recent trends and future direction of dental research
in the digital era. Int. J. Environ. Res. Public Health 2020, 17, 1987. [CrossRef]
2. Shan, T.; Tay, F.; Gu, L. Application of artificial intelligence in dentistry. J. Dent. Res. 2021, 100, 232–244. [CrossRef]
3. Joda, T.; Yeung, A.; Hung, K.; Zitzmann, N.; Bornstein, M. Disruptive Innovation in Dentistry: What It Is and What Could Be
Next. J. Dent. Res. 2020, 0022034520978774.
4. Park, W.J.; Park, J.-B. History and application of artificial neural networks in dentistry. Eur. J. Dent. 2018, 12, 594. [CrossRef]
5. Ganguly, B.; Chaudhuri, S.; Biswas, S.; Dey, D.; Munshi, S.; Chatterjee, B.; Dalai, S.; Chakravorti, S. Wavelet kernel-based
convolutional neural network for localization of partial discharge sources within a power apparatus. IEEE Trans. Ind. Inform.
2020, 17, 1831–1841. [CrossRef]
6. Moradi, M.J.; Hariri-Ardebili, M.A. Developing a Library of Shear Walls Database and the Neural Network Based Predictive
Meta-Model. Appl. Sci. 2019, 9, 2562. [CrossRef]
7. Roshani, M.; Phan, G.T.; Ali, P.J.M.; Roshani, G.H.; Hanus, R.; Duong, T.; Corniani, E.; Nazemi, E.; Kalmoun, E.M. Evaluation of
flow pattern recognition and void fraction measurement in two phase flow independent of oil pipeline’s scale layer thickness.
Alex. Eng. J. 2021, 60, 1955–1966. [CrossRef]
8. Ren, R.; Luo, H.; Su, C.; Yao, Y.; Liao, W. Machine learning in dental, oral and craniofacial imaging: A review of recent progress.
PeerJ 2021, 9, e11451. [CrossRef] [PubMed]
9. Alafif, T.; Tehame, A.M.; Bajaba, S.; Barnawi, A.; Zia, S. Machine and Deep Learning towards COVID-19 Diagnosis and Treatment:
Survey, Challenges, and Future Directions. Int. J. Environ. Res. Public Health 2021, 18, 1117. [CrossRef]
10. Miyazaki, T.; Hotta, Y. CAD/CAM systems available for the fabrication of crown and bridge restorations. Aust. Dent. J. 2011, 56,
97–106. [CrossRef] [PubMed]
11. Bernauer, S.A.; Müller, J.; Zitzmann, N.U.; Joda, T. Influence of Preparation Design, Marginal Gingiva Location, and Tooth
Morphology on the Accuracy of Digital Impressions for Full-Crown Restorations: An In Vitro Investigation. J. Clin. Med. 2020, 9,
3984. [CrossRef]
Sensors 2021, 21, 6628 11 of 11

12. Gintaute, A.; Straface, A.; Zitzmann, N.U.; Joda, T. Die Modellgussprothese 2.0: Digital von A bis Z. Swiss. Dent. J. 2020, 130,
229–235. [PubMed]
13. Saravi, B.; Vollmer, A.; Hartmann, M.; Lang, G.; Kohal, R.-J.; Boeker, M.; Patzelt, S. Clinical performance of CAD/CAM All-
ceramic tooth-supported fixed dental prostheses: A systematic review and meta-analysis. Materials 2021, 14, 2672. [CrossRef]
[PubMed]
14. Currie, G. Intelligent imaging: Anatomy of machine learning and deep learning. J. Nucl. Med. Technol. 2019, 47, 273–281.
[CrossRef]
15. Albus, J.S. Outline for a theory of intelligence. IEEE Trans. Syst. Man Cybern. 1991, 21, 473–509. [CrossRef]
16. Meyer, P.; Noblet, V.; Mazzara, C.; Lallement, A. Survey on deep learning for radiotherapy. Comput. Biol. Med. 2018, 98, 126–146.
[CrossRef]
17. Hung, K.; Montalvao, C.; Tanaka, R.; Kawai, T.; Bornstein, M.M. The use and performance of artificial intelligence applications in
dental and maxillofacial radiology: A systematic review. Dentomaxillofacial Radiol. 2020, 49, 20190107. [CrossRef] [PubMed]
18. Grischke, J.; Johannsmeier, L.; Eich, L.; Griga, L.; Haddadin, S. Dentronics: Towards robotics and artificial intelligence in dentistry.
Dent. Mater. Off. Publ. Acad. Dent. Mater. 2020, 36, 765–778. [CrossRef]
19. Lee, J.-H.; Jeong, S.-N. Efficacy of deep convolutional neural network algorithm for the identification and classification of dental
implant systems, using panoramic and periapical radiographs: A pilot study. Medicine 2020, 99. [CrossRef]
20. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G.; Group, P. Preferred reporting items for systematic reviews and meta-analyses:
The PRISMA statement. PLoS Med. 2009, 6, e1000097. [CrossRef]
21. Lerner, H.; Mouhyi, J.; Admakin, O.; Mangano, F. Artificial intelligence in fixed implant prosthodontics: A retrospective study of
106 implant-supported monolithic zirconia crowns inserted in the posterior jaws of 90 patients. BMC Oral Health 2020, 20, 1–16.
[CrossRef] [PubMed]
22. Yamaguchi, S.; Lee, C.; Karaer, O.; Ban, S.; Mine, A.; Imazato, S. Predicting the debonding of CAD/CAM composite resin crowns
with AI. J. Dent. Res. 2019, 98, 1234–1238. [CrossRef] [PubMed]
23. Lee, J.-H.; Kim, D.-H.; Jeong, S.-N.; Choi, S.-H. Detection and diagnosis of dental caries using a deep learning-based convolutional
neural network algorithm. J. Dent. 2018a, 77, 106–111. [CrossRef]
24. Lee, J.-H.; Kim, D.-h.; Jeong, S.-N.; Choi, S.-H. Diagnosis and prediction of periodontally compromised teeth using a deep
learning-based convolutional neural network algorithm. J. Periodontal Implant. Sci. 2018, 48, 114. [CrossRef]
25. Raith, S.; Vogel, E.P.; Anees, N.; Keul, C.; Güth, J.-F.; Edelhoff, D.; Fischer, H. Artificial Neural Networks as a powerful numerical
tool to classify specific features of a tooth based on 3D scan data. Comput. Biol. Med. 2017, 80, 65–76. [CrossRef]
26. Wei, J.; Peng, M.; Li, Q.; Wang, Y. Evaluation of a Novel Computer Color Matching System Based on the Improved Back-
Propagation Neural Network Model. J. Prosthodont. 2018, 27, 775–783. [CrossRef] [PubMed]
27. Zitzmann, N.U.; Krastl, G.; Hecker, H.; Walter, C.; Waltimo, T.; Weiger, R. Strategic considerations in treatment planning: Deciding
when to treat, extract, or replace a questionable tooth. J. Prosthet. Dent. 2010, 104, 80–91. [CrossRef]
28. Lee, J.-H.; Kim, D.-H.; Jeong, S.-N. Diagnosis of cystic lesions using panoramic and cone beam computed tomographic images
based on deep learning neural network. Oral Dis. 2020, 26, 152–158. [CrossRef]
29. Hiraiwa, T.; Ariji, Y.; Fukuda, M.; Kise, Y.; Nakata, K.; Katsumata, A.; Fujita, H.; Ariji, E. A deep-learning artificial intelligence
system for assessment of root morphology of the mandibular first molar on panoramic radiography. Dentomaxillofacial Radiol.
2019, 48, 20180218. [CrossRef]
30. Kurt Bayrakdar, S.; Orhan, K.; Bayrakdar, I.S.; Bilgir, E.; Ezhov, M.; Gusarev, M.; Shumilov, E. A deep learning approach for dental
implant planning in cone-beam computed tomography images. BMC Med. Imaging 2021, 21, 86. [CrossRef]
31. Chen, Q.; Wu, J.; Li, S.; Lyu, P.; Wang, Y.; Li, M. An ontology-driven, case-based clinical decision support model for removable
partial denture design. Sci. Rep. 2016, 6, 1–8. [CrossRef] [PubMed]
32. Joda, T.; Gallucci, G.; Wismeijer, D.; Zitzmann, N. Augmented and virtual reality in dental medicine: A systematic review. Comput.
Biol. Med. 2019, 108, 93–100. [CrossRef] [PubMed]
33. Zitzmann, N.U.; Matthisson, L.; Ohla, H.; Joda, T. Digital undergraduate education in dentistry: A systematic review. Int. J.
Environ. Res. Public Health 2020, 17, 3269. [CrossRef] [PubMed]
34. Van der Meer, W.J.; Andriessen, F.S.; Wismeijer, D.; Ren, Y. Application of intra-oral dental scanners in the digital workflow of
implantology. PLoS ONE 2012, 7, e43312. [CrossRef] [PubMed]