|Year : 2023 | Volume
| Issue : 2 | Page : 134-141
Artificial intelligence in endodontics: A narrative review
Parvathi Sudeep, Paras M Gehlot, Brindha Murali, Annapoorna B Mariswamy
Department of Conservative Dentistry and Endodontics, JSS Dental College and Hospital, JSS Academy of Higher Education and Research (JSSAHER), JSS Medical Institutions Campus, Mysuru, Karnataka, India
|Date of Submission||06-Dec-2022|
|Date of Decision||12-Jan-2023|
|Date of Acceptance||13-Jan-2023|
|Date of Web Publication||28-Apr-2023|
Dr. Paras M Gehlot
Department of Conservative Dentistry and Endodontics, JSS Dental College and Hospital, JSS Academy of Higher Education and Research (JSSAHER), JSS Medical Institutions Campus, Sri Shivarathreeshwara Nagara, Mysuru 570015, Karnataka
Source of Support: None, Conflict of Interest: None
Aim: With the help of developments in artificial intelligence (AI), picture archiving systems, and computer-aided diagnostic systems, dentists have been able to augment the quality of treatment and ensure a favorable outcome, by improving and facilitating the delivery of appropriate dental care. There has been a breakthrough in designing the diagnosis, treatment plans, and predicting prognoses recently, which has helped to explore newer options for better treatment. Materials and Methods: A literature search was conducted using MeSH terms in a variety of databases, including PubMed, Cochrane, Scopus, and Web of Science, to gather information on “Artificial intelligence (AI) in endodontics.” Unpublished data, literature written in other languages, and articles with only abstracts were discarded. Forty-one relevant articles were included. Results: Since there were not many papers referring to AI in endodontics, papers published relating to AI in dentistry were also referred. The search showed that the use of AI in dentistry, specifically in endodontics, has enormous promise. Although useful, AI has its disadvantages as well as the need for long-term studies. Conclusion: AI, consisting of a sequence of algorithms, work on a concept that mimics the human brain and thinking. AI in endodontics has been used widely in locating apical foramina, identifying periapical pathologies, diagnosis of vertical root fractures, evaluating the outcome of regenerative procedures and retreatments, and assessment of root morphologies and difficulties associated with canal preparations. Being a potential game changer and beginning something called a “fourth industrial revolution,” AI has what it takes to revolutionize endodontics with time.
Keywords: Artificial Intelligence, Deep Learning, Endodontics, Machine Learning, Neural Networks
|How to cite this article:|
Sudeep P, Gehlot PM, Murali B, Mariswamy AB. Artificial intelligence in endodontics: A narrative review. J Int Oral Health 2023;15:134-41
|How to cite this URL:|
Sudeep P, Gehlot PM, Murali B, Mariswamy AB. Artificial intelligence in endodontics: A narrative review. J Int Oral Health [serial online] 2023 [cited 2023 Jun 4];15:134-41. Available from: https://www.jioh.org/text.asp?2023/15/2/134/375370
| Introduction|| |
Digitalization in dentistry has increased remarkably over the last two decades. In most developing countries, the shortage of medical and dental professionals demands the need for technology, especially artificial intelligence (AI) software. This can reduce expenses, time, the need for human knowledge, and the number of medical errors. A model designed to work almost identically to the human brain has been an enigma for a lot of researchers and scientists over the years. Their constant effort and hard work resulted in the evolution of “Artificial Intelligence.” The term AI, explains how, through technology, a machine or software, may replicate human intellect to do particular tasks.
In 1955, John McCarthy coined a term which explained the possibility of machines to carry out activities that can be encompassed within “intelligent” tasks. In the year 1956, McCarthy, also widely recognized as the “Father of Artificial intelligence,” organized a conference in Dartmouth, which paved the way for the extensive research on AI, from 1950s to 1970’s.
AI can be described as “a field of science and engineering concerned with the computational understanding of what is commonly called intelligent behavior, and with the creation of artefacts that exhibit such behavior.”
Today, dentistry has abundant data which are acquired from a patient at each step of the treatment process. This poses a challenge for obtaining, examining, and utilizing the knowledge gained. The vast amount of acquired knowledge can be deemed necessary to solve complex clinical problems. The evolution of AI will likely not replace dentists, but would be a perfect addition to accentuate their expertise and support them. This can indeed help a dentist to adapt to a new level of accuracy in diagnosis and prediction of outcomes, thus, rendering better patient experience and treatment success.
The lack of standardization of treatment protocols can also pose a problem when newer technologies are used. Hence, the use of AI can also be beneficial in understanding and identifying conditions wherein the said technology can be applied.
To understand how AI can be utilized, key aspects such as machine learning (ML), neural networks (NN), and genetic algorithms have to be acknowledged. Previous studies on the use of AI in dentistry explains how AI can be an advantage in the various aspects of patient management and treatment strategies.
The purpose of this narrative review was to understand about the application of AI in endodontics and the effects that any future advancements in this area might have in this field of dentistry. In light of the paucity of research in this area, the review here is based on articles taken from internationally published research literature and review articles, based on a specific set of key words which emphasized on the published information about AI, its various applications in the field of Endodontics and regarding its limitations as well. Computer databases were utilized for identifying the articles with the required information.
| Materials and Methods|| |
The databases of PubMed/Medline and other sources which related with AI in endodontics, were thoroughly searched from August 2021 to September 2022. The following MeSH and keywords were used in the search: “Artificial intelligence, Deep learning, Endodontics, Machine learning, and Neural networks.”
The PICOS principles were followed while choosing the inclusion criterion. The articles searched and included were the following:
- - All English-language papers with full-text abstracts
- - Articles discussing the relationship between AI and applications in endodontics
- - Articles published since the use of AI in dentistry
- - Studies (in vitro and in vivo) that explored the use of AI in endodontics
The exclusion criterion included case series/reports, articles written in languages other than English, unpublished data, and unrelated studies. Dual publications were also turned down. Possible outcomes were presented once the data were examined and compared with each particular paper under debate. On the retrieved data, no statistical analysis was performed [Figure 1].
| Results|| |
Study recognition and preference
The initial electronic and manual search yielded 251 results the PubMed, Scopus, Cochrane and Web of Science databases (2021–2022). On initial screening and application of inclusion criteria 62 articles were chosen based on name, abstract, and whole text. Finally, the narrative reviews, comprehensive reviews, umbrella reviews, and clinical trials that met the criteria were chosen.
Characteristics of included studies
Randomized controlled trials, in vitro and in vivo research, and reviews were all included. Forty-one papers that met the requirements for inclusion were included in our review. Out of these, 22 papers covered AI in dentistry and 19 papers examined the applications of AI in endodontics.
Assessment and quality of proof in studies included
The literature that is currently available in the research (reviews, in vitro, and in vivo) all agree about the beneficial impact of application of AI in the various fields of dentistry. On this topic though, there were no meta-analyses and the number of clinical trials appears to be negligible. Future clinical research is crucial to understand how AI can be applied in every aspect of dentistry and especially, endodontics. Almost all studies employ the best practices, and all relevant findings are presented in a consistent, verified, and reliable manner.
| Discussion|| |
AI methodologies used in dentistry
The composition of the technology used in AI consists of a neural network pattern almost identical to human brains. This neural design imitates human thinking, and is composed of firmly interconnected neurons. To solve specific problems, these work together as a “data-processing system.” There has been amazing developments in the application of neural networks in dentistry over the years.
Conventional computer programs such as “expert systems” use complicated formulas and numerical models to accomplish computing and achieve an array of plans formulated on the given programming patterns. At present, AI research has entered a distinct field of study known as “machine learning (ML).” There are other key areas of research included in AI as well [Figure 2].
Machine learning is an AI subset, that uses algorithms to anticipate outcomes based on a dataset. These algorithms are used to discover data’s inherent statistical patterns and structures. Machine learning aims to make it easier for machines to learn from and solve issues from large data sets without the need for human intervention. For training, ML makes use of computational tools and data (experience). It does so to assess the data presented as input and to process the knowledge gained from former experiences. “Experience gathering” or “active learning” is at the heart of machine learning. In practice, this means that computers learn from their incoming data and improve their properties by learning from their mistakes, all without the requirement for complex programming or the building of a numerical model.
In a neural network, the number of neurons, levels or eras; membership function selections in fuzzy logic; the size of the population, selection methods, rate of mutations, and crossovers in optimization techniques; and hybrid techniques that use fuzzy logic or neural network or both inevitably necessitate parametric tweaking related to the objective scheme. Various ML models, such as the “Genetic Algorithm (GA),” “Artificial Neural Network (ANN),” and “Support Vector Machine (SVM),” may “learn” and “be trained” from provided data to perform a variety of tasks.
A prominent sort of ML model is neural networks, which are made up of a series of algorithms that calculate signals using artificial neurons. Its goal is to develop neural networks that mimic the human brain.
Artificial neural networks are a collection of a configurable number of “artificial neurons or nodes” that are connected in a systematic layering structure that includes an input sequence, one or maybe more hidden units, and an output vector. With the exception of the input nodes, each neuron receives numerous weighted inputs and produces an output that is usually an arbitrary expression of the inputs. By altering these weights regularly, a neural network “learns.”
Their capacity to apply what they’ve learned from previous representative cases, analyze non-linear data, deal with unspecific data, and generalize; allowing the model to be exposed to new information, has made them a very appealing investigative tool. They have been employed in clinical diagnosis, radiology and histopathology image processing, intensive care data interpretation, and waveform analysis.
“Convolutional neural networks (CNNs)” for image classifications and “dilated convolutional neural networks (DCNNs)” for sematic scene partitions are two types of ANN that have recently gained popularity. The two primary types of CNNs used for volumetric prediction in particular are “Tiramisu” and DCNNs. “U-net” and other Tiramisu-based models are particularly good at estimating dosage ratios that are spatially compatible with anatomy, such as the prostate dose volume. DCNNs also utilize the complexities that bypass information during the encoding process to help broaden their perspectives and visual field.
DCNNs are likely to become more common in the prediction of the amount of exposure for head and neck intensity modulated radiation therapy as they can help estimate the dose that is mobile with respect to anatomy, such as in patients with head and neck cancer.
Deep learning is a part of ML that process information using a deep neural network with multiple computational layers. To enhance feature identification, deep learning is utilized to develop a computational model that detects patterns automatically. They employ simple features such as line, edge, and texture to study complex systems, pathologies, or entire organs., The difference between deep learning and NNs is similar to that between feed forward NNs and feed backward NNs. Deep learning has a more complicated technique of connecting layers, as well as a higher number of neurons than other networks to demonstrate sophisticated programs, more processing capacity to train, and a preprogrammed character recognition feature.
Clinical decision support systems (CDSS)
“Any computer system that aids healthcare providers in making clinical choices by handling clinical data or medical knowledge” is referred to as a CDSS. The inference engine (IE), knowledge base (KB), explanation module, and working memory are the four essential components of most CDSS. Any such system’s primary component is the IE, which stores information about the patient and draws conclusions about specific illnesses from it. The knowledge used by IE is represented by the knowledge base and tools that have been built to aid in the accumulation and elaboration of this knowledge. The working memory is the database where the acquired data are saved or may take the shape of a signal. The final component, the explanation module, is in charge of justifying the IE’s conclusions when applying the KB’s information to patient data and records in the working memory.
The branch of logic known as fuzzy logic recognizes and proves that everything is a matter of degree. It offers quite useful explanation for a hazy problem. Fuzzy logic is a very simple ML technique that is also flexible to use.
The following are the four major components of the fuzzy logic system:
(i) “Rule Basse,” (ii) “Fuzzification,” (iii) “Inference Engine,” and (iv) “Defuzzification.”
Fuzzy logic aims to emulate human reasoning abilities when dealing with unclear issues.
Evolutionary computation is a disputed dynamic programming strategy for general-purpose problems based on the generally accepted “neo-Darwinian” perspective, which incorporates basic Darwinian evolutionary theory, Weismann’s selectionism, and Mendel’s genetics. The genetic algorithm is the most extensively used evolutionary algorithm. It uses various methods including mutation, inheritance, selection, and crossover to find a better remedy. The key advantage of genetic algorithms over traditional methods is that they are based on conflict resolutions rather than numerical correlations, as used by traditional methods. It is simple to use despite being a powerful upgrading device, because it is based on normal standards.
Hybrid intelligent systems
Each AI technique has its own set of advantages and disadvantages. Combining systems enable common sense to be accommodated, knowledge to be extracted from original assets, “human-like” cognitive mechanisms to be used, ambiguity and inaccuracy to be dealt with, and learning to keep up with the rapidly evolving and unfamiliar environment. Developing fuzzy systems for ANNs and vice versa, and Genetic Algorithms for autonomously instructing and designing NNA structures are just a few of the hybrid systems accessible.
Endodontics and applications of AI [[Figure 3]]
Identification and diagnosis
AI has been used to detect and diagnose a wide range of physiological and pathological abnormalities in the teeth. AI has also been shown in studies to be an effective tool for understanding and spotting anomalies, as well as for formulating the required treatment strategies.
Radiographically, periapical pathosis can be detected as periapical radiolucencies. On periapical radiographs, panoramic radiographs, and cone-beam computed tomography scans (CBCT), apical periodontitis is frequently discovered as an unintentional finding. CBCT produces high-resolution three-dimensional (3D) images that are free of the deformation and superimposition of bone and dental structures that might occur with traditional radiography. A dentist knows that it is always crucial to have a dependable instrument that can assist in making informed selections and treatment planning. A study looked at how an AI-based CNN model may be used to detect periapical lesions on radiographs, and found that the model performed well, with high sensibility and intermediate precision. For dentists, this approach could be very useful in identifying and detecting periapical lesions.
Another model’s accuracy in recognizing periapical diseases on CBCT images was astounding, with a 92.8% accuracy rate. An AI-based model was utilized in another investigation to detect the existence of a periapical lesion. There were unexpected findings, with a mean accuracy of roughly 77.2%, and the authors stated that the reference method was better than this approach, but that the results may be improved in future work utilizing optimization techniques.
Setzer et al. concluded that a DL algorithm trained in a constrained CBCT setting produced high lesion detection accuracy. Enhanced versions of AI may improve overall voxel-matching accuracy, thus, revealing an accuracy of 93% for this study.
Endres et al. found that the rate of detection of periapical radiolucencies by 24 oral and maxillofacial surgeons was similar to using a DL algorithm model. The experiments revealed that adopting AI systems can lessen the differences between examiners and bias.
Poswar et al. analyzed the variations in gene expression between a periapical cyst and a periapical granuloma using a “multilayer perceptron neural network” for gene categorization. The research had its own set of constraints.
Zheng et al. compared a morphologically restricted Dense U-Net with known clinical image analysis approaches in terms of lesion diagnosis efficiency and dice coefficient indices of a multilabel segmentation. Despite the small size of the sample, the researchers found that the novel deep learning strategy increased CBCT segmentation and the precision of pathological identification.
“Vertical root fractures (VRFs)” are a very uncommon occurrence in endodontically treated teeth. VRFs are found in 3.7%–30.8% of teeth that have received endodontic therapy, according to studies. Determining VRFs on radiographs is difficult and may necessitate the use of more advanced technology.
Johari et al. performed research on the creation of an AI-based model utilizing a PNN framework for detecting VRFs on both intact and endodontically treated teeth using periapical radiographs and CBCT images. In comparison to periapical radiographs, the model demonstrated to be quite effective in diagnosing VRFs on CBCT images, with an accuracy of 96.6%.
CNNs, according to Fukuda et al., could be a viable approach for detecting and measuring VRFs on panoramic radiographs. In a study by Kositbowornchai et al., who employed a probabilistic neural network design to determine if a tooth root was sound or had a vertical root fracture, the accuracy was 95.7%.
Root canal and root morphology
For a successful non-surgical root canal therapy, dentists must first understand and identify root canal anatomy and root morphologies. AI can assist in recognizing any morphological irregularities as well as identifying errors in locating new canals.
On panoramic radiographs, Hiraiwa et al. investigated the diagnostic efficacy of a DL system for categorization of the root morphology of mandibular first molars. On CBCT images, distal roots were investigated for the occurrence of a single or additional root. For determining whether distal roots were solitary or had supplementary roots, the DL system exhibited a diagnostic reliability of 86.9%.
Lahoud et al. used the CNN technique to automate 3-dimensional tooth segmentation. In a rapid, efficient, and reliable clinical benchmark, the authors analyzed 433 CBCT radiographic segmentations of teeth and found that AI performed as well as a human operator but at a considerably quicker pace.,
Working length determinations
The selection of the correct working length (WL) is critical to ensuring successful root canal treatment outcomes. Instrumentation beyond the apical foramen, flare-ups, periapical foreign body reactions, and poor microbiological control are all common consequences of inadequate WL determination. Finding the apical foramen and estimating the WL can be done in a variety of ways, including radiography, digital tactile sense, and patient’s responses to a file or paper point. The use of digital technology has demonstrated both benefits in terms of finding the apical foramen and drawbacks in terms of mistakes. As a result, investigations on the use of ANN to estimate the correct WL of teeth have been done.
In a human cadaver model, Saghiri et al. used an artificial neural network (ANN) method to determine the working length and demonstrated an outstanding accuracy of 96%, which is higher than the accuracy of expert endodontists (76%). These findings matched those of another study by the same author and his associates, who used feature extraction processes from radiographs before processing the data with an ANN. The study found that finding the minor apical foramen was 93% accurate.
CBS stands for “case-based reasoning,” which defines the practice of coming up with solutions to queries and doubts based on previous encounters with similar challenges. Campo et al. used CBS to predict the results of nonsurgical root canal retreatment, as well as the associated risks and benefits. The system gave the data on whether or not retreatment should be conducted, and also provided the statistics on the efficiency of the procedure done earlier, the risks involved, and the recall periods. The system’s merit is that it may be able to properly forecast the retreatment outcome. But the technique had a drawback of relying too much on the data’s information, and not analyzing possibilities from every perspective.
Regenerative endodontic procedures
Using the neuro-fuzzy inference technique, Bindal et al. examined dental pulp stem cells in multiple rejuvenation treatment modalities. This approach predicted the result of a simulated clinical scenario by assessing the stem cell vitality after bacterial lipopolysaccharide treatment. Throughout several regenerative regimens, the neuro-fuzzy interpretation method was employed to predict cell viability, following microbial infection. The researchers used neuro-fuzzy inferencing, which could be adjustable, to test the outcome’s accuracy in forecasting stem cell survival after pathogenic invasion.,
Shortcomings and gaps in AI and endodontics
Though AI has been found to be beneficial in a lot of aspects, there have been downfalls associated with it, due to which it has not been widely implemented in endodontics. Some of the shortcomings and disadvantages associated include the following:,
- i. The health care system requires a standard programming technology for organizing appointments, managing patients and for periodic recall. This system also has to be updated regularly to keep up to the changes occurring in the healthcare sector.
- ii. The cost of setting up an AI based system in independent clinics would not be feasible.
- iii. The availability of patient data using the individual health records can prove to be useful in informing the dentists about adverse reactions in respect to the use of certain drugs and/or the adjustments required in the treatment protocol. But this can also be a disadvantage, as misuse of information can also occur through the AI channels.
- iv. There is no protocol that has been devised yet to provide a precise diagnosis, that would affect the prediction of outcomes or determination of the prognosis. The clarity of this diagnosis would be based on the clinical findings.
- v. Dynamic navigations in endodontic surgery can be improved with the help of AI. Very minimal number of studies look into these techniques and it is highly recommended that further studies be done to check and assess robotically guided placement in endodontics.
- vi. The complexity of the applications can be misinterpreted by individuals and dentists who have not been trained in AI.
- vii. Data in endodontics can be often used for both education and evaluation, leading to “data snooping bias.”
- viii. The results of AI in endodontics are not readily significant as AI can give a single outcome as an answer, where multiple outcomes could be present.
The limitation of this review is the constrained search period (August 2021 to September 2022). Also, papers with abstracts only and published in foreign languages were not included. It is interesting to note that, a few more articles related to AI in endodontics have been added to the literature lately.
| Conclusion|| |
With inter-disciplinary dentistry becoming more popular in the coming decades, it is indeed important to assess the worth of AI and its applications in endodontics. The fate of Endodontics, which is a highly researched subject, can be changed by the incorporation of modified bioengineering components like AI. The impact of AI in integrating the various areas of knowledge can prove to be beneficial in upgrading patient care. The combination of conventional methodologies with AI technology can help in minimizing the potential errors in interpretation and can indefinitely, be adopted into routine clinical practice with ease.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
PS and PG conceived the idea and study design. PS performed the literature search and the extraction of data. PS drafted the article. PG, BM and AM reviewed the article.
Ethical policy and Institutional Review board statement
Patient declaration of consent
Data availability statement
| References|| |
Machoy ME, Szyszka-Sommerfeld L, Vegh A, Gedrange T, Woźniak K, Woźniak K The ways of using machine learning in dentistry. Adv Clin Exp Med 2020;29:375-84.
Tandon D, Rajawat J Present and future of artificial intelligence in dentistry. J Oral Bio Craniofacial Res 2020;10:391-6.
Khanagar SB, Ehaideb AA, Maganur PC, Vishwanathaiah S, Patil S, Baeshen HA, et al
. Developments, application, and performance of artificial intelligence in dentistry—A systematic review. J Dent Sci 2021;19:508-22.
Rajaraman V John McCarthy father of artificial intelligence. Reson. 2014;19:198-207.
National Research Council. Funding a Revolution: Government Support for Computing Research. Washington, DC: The National Academies Press, 1999.
Shapiro SC Artificial intelligence. In: Shapiro SC, editor. Encyclopedia of Artificial Intelligence. Vol. 1. New York: Wiley, 1992.
Ramesh AN, Kambhampati C, Monson JRT, Drew PJ Artificial intelligence in medicine. Ann R Coll Surg Engl 2004;86:334-8.
Brickley MR, Shepherd JP, Armstrong RA Neural networks: A new technique for development of decision support systems in dentistry. J Dent 1998;26:305-9.
Tripathy M, Maheshwari RP, Verma HK Power transformer differential protection based on optimal probabilistic neural network. IEEE Trans Power Del 2010;25:102-12.
Ren R, Luo H, Su C, Yao Y, Liao W Machine learning in dental, oral and craniofacial imaging: A review of recent progress. PeerJ 11459:1.
Schwendicke F, Samek W, Krois J Artificial intelligence in dentistry: Chances and challenges. J Dent Res 2020;99:769-74.
Erickson BJ, Korfiatis P, Akkus Z, Kline TL Machine learning for medical imaging. Radiographics 2017;37:505-15.
Kareem SA, Pozos-Parra P, Wilson N An application of belief merging for the diagnosis of oral cancer. Appl Soft Comput J 2017;61:1105-12.
Tan MS, Tan JW, Chang S-W, Yap HJ, Abdul Kareem S, Zain RB A genetic programming approach to oral cancer prognosis. PeerJ 2016;4:e24822482.
Nayak GS, Kamath S, Pai KM, Sarkar A, Ray S, Kurien J, et al
. Principal component analysis and artificial neural network analysis of oral tissue fluorescence spectra: Classification of normal premalignant and malignant pathological conditions. Biopolymers 2006;82:152-66.
Shilpi S Artificial intelligence in dentistry: The current concepts and a peek into the future. Int J Contemp Med 2019;6:L5-9.
Kearney V, Chan JW, Valdes G, Solberg TD, Yom SS The application of artificial intelligence in the IMRT planning process for head and neck cancer. Oral Oncol 2018;87:111-6.
Akst J A primer: Artificial intelligence versus neural networks. Inspiring Innovation: The Scientist Exploring Life 2019. p. 65802.
Rabun al JR, Dorado J Artificial neural networks in real-life applications. IGI Global: Hershey 2005;1:166-346.
Yaji A, Prasad S, Pai A Artificial intelligence in dento-maxillofacial radiology. Acta Sci Dent Sci 2019;3:116-21.
Abiodun OI, Jantan A, Omolara AE, Dada KV, Mohamed NAE, Arshad H State-of-the-art in artificial neural network applications: A survey. Heliyon 2018;4:e00938.
Khalifa M Clinical decision support: Strategies for success. Proc Comput Sci 2014;37:422-7.
Zhang B, Wu Y, Lu J, Du K Evolutionary computation and its applications in neural and fuzzy systems. Appl Comput Intell Soft Comput2011;2011:1-20.
Tripathi P, Malathy C, Prabhakaran M Genetic algorithms based approach for dental caries detection using back propagation neural network. Int J Recent Technol Eng 2019;8:2277-3878.
Umer F, Habib S Critical analysis of artificial intelligence in endodontics: A scoping review. J Endod 2022;48:152-60.
AAmerican Association of Endodontists, American Academy of Oral and Maxillofacial Radiology. Use of cone-beam computed tomography in endodontics Joint Position Statement of the American Association of Endodontists and the American Academy of Oral and Maxillofacial Radiology. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2011;111:234-7.
Ekert T, Krois J, Meinhold L, Elhennawy K, Emara R, Golla T, et al
. Deep learning for the radiographic detection of apical lesions. J Endod 2019;45:917-922.e5.
Orhan K, Bayrakdar IS, Ezhov M, Kravtsov A, Özyürek T Evaluation of artificial intelligence for detecting periapical pathosis on cone-beam computed tomography scans. Int Endod J 2020;53:680-9.
Mahmoud YE, Labib SS, Hoda MO Clinical prediction of teeth periapical lesion based on machine learning techniques—An experimental study. Proceedings of Second International Conference on Digital Information Processing, Data Mining, and Wireless Communications (DIPDMWC 2015), Dubai, UAE, 2015.
Setzer FC, Shi KJ, Zhang Z, Yan H, Yoon H, Mupparapu M, et al
. Artificial intelligence for the computer-aided detection of periapical lesions in cone-beam computed tomographic images. J Endod 2020;46:987-93.
Endres MG, Hillen F, Salloumis M, Sedaghat AR, Niehues SM, Quatela O, et al
. Development of a deep learning algorithm for periapical disease detection in dental radiographs. Diagnostics (Basel) 2020;10:430.
Poswar Fde O, Farias LC, Fraga CA, Bambirra W, Brito-Júnior M, Sousa-Neto MD, et al
. Bioinformatics, interaction network analysis, and neural networks to characterize gene expression of radicular cyst and periapical granuloma. J Endod 2015;41:877-83.
Zheng Z, Yan H, Setzer FC, Shi KJ, Mupparapu M, Li J Anatomically constrained deep learning for automating dental CBCT segmentation and lesion detection. IEEE Trans Autom Sci Eng 2021;18:603-14.
Fuss Z, Lustig J, Katz A, Tamse A An evaluation of endodontically treated vertical root fractured teeth: Impact of operative procedures. J Endod 2001;27:46-8.
Johari M, Esmaeili F, Andalib A, Garjani S, Saberkari H Detection of vertical root fractures in intact and endodontically treated premolar teeth by designing a probabilistic neural network: An ex vivo study. Dentomaxillofac Radiol 2017;46:20160107.
Fukuda M, Inamoto K, Shibata N, Ariji Y, Yanashita Y, Kutsuna S, et al
. Evaluation of an artificial intelligence system for detecting vertical root fracture on panoramic radiography. Oral Radiol 2019;36:337-343.
Kositbowornchai S, Plermkamon S, Tangkosol T Performance of an artificial neural network for vertical root fracture detection: An ex vivo study. Dent Traumatol 2013;29:151-5.
Hiraiwa T, Ariji Y, Fukuda M, Kise Y, Nakata K, Katsumata A, et al
. A deep-learning artificial intelligence system for assessment of root morphology of the mandibular first molar on panoramic radiography. Dentomaxillofac Radiol 2019;48:20180218.
Lahoud P, EzEldeen M, Beznik T, Willems H, Leite A, Van Gerven A, et al
. Artificial intelligence for fast and accurate 3-dimensional tooth segmentation on cone-beam computed tomography. J Endod 2021;47:827-35.
Aminoshariae A, Kulild J, Nagendrababu V Artificial intelligence in endodontics: Current applications and future directions. J Endod 2021;47:1352-7.
Saghiri MA, Garcia-Godoy F, Gutmann JL, Lotfi M, Asgar K The reliability of artificial neural network in locating minor apical foramen: A cadaver study. J Endod 2012;38:1130-4.
Seidberg B, Alibrandi B, Fine H, Logue B Clinical investigation of measuring working lengths of root canals with an electronic device and with digital-tactile sense. J Am Dent Assoc 1975;90:379-87.
Saghiri MA, Asgar K, Boukani KK, Lotfi M, Aghili H, Delvarani A, et al
. A new approach for locating the minor apical foramen using an artificial neural network. Int Endod J2012;45:257-65.
Campo L, Aliaga IJ, De Paz JF, García AE, Bajo J, Villarubia G, et al
. Retreatment predictions in odontology by means of CBR systems. Comput Intell Neurosci 2016;2016:7485250.
Bindal P, Bindal U, Lin CW, Kasim NHA, Ramasamy TSA, Dabbagh A, et al
. Neuro-fuzzy method for predicting the viability of stem cells treated at different time-concentration conditions. Technol Health Care 2017;25:1041-51.
Chen YW, Stanley K, Att W Artificial intelligence in dentistry: Current applications and future perspectives. Quintessence Int 2020;51:248-57.
[Figure 1], [Figure 2], [Figure 3]