Artificial Intelligence and Dentistry: The Future

ABSTRACT

Artificial intelligence (AI) is the ability of machines to perform tasks that normally require human intelligence. Although it was first conceptualized in the 1950s, it has developed rapidly in recent times. AI can undertake complex predictions and make decisions due to its ability to mimic human intelligence. Its popularity has grown exponentially in the health sector including dentistry. It is useful in all dental disciplines, including endodontics, oral medicine and radiology, periodontics, prosthodontics, oral pathology, and forensic odontology. This review describes the history, types, and uses of AI across different disciplines. This will help dental professionals to understand the utility of AI and improve their efficiency toward better patient care.

INTRODUCTION

Artificial intelligence (AI) is defined as “The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision making, and translation between languages. It performs tasks that require human intelligence by learning from human expertise.[1] The use of AI is increasing in many fields of industry such as finance, automobiles and robotics, and healthcare. The use of AI in health care is increasing exponentially. It can mimic human intelligence and perform decision-making in the healthcare sector.[2] AI is emerging as a valuable tool in dentistry both in diagnosis and treatment planning AI models, such as artificial neural networks (ANNs) and convolutional neural networks (CNNs), have shown great promise in dentistry. The role of AI routinely in dentistry is yet to emerge. However, it has great potential to revolutionize the way dentistry is practiced. Therefore, this update aims to put forth the developments in all\ fields of dentistry with respect to AI and further dwell upon the implications of AI on the future of dentistry.

HISTORY OF ARTIFICIAL INTELLIGENCE

The concept of AI can be traced back to the ancient Indian text, the Vedas, which were written between 1500 BC and 500 BC. The Vedas contain various references to machines and robots that can perform human-like tasks. For example, the Rig Veda describes a chariot that is pulled by horses made of metal, whereas the Yajur Veda describes machines that can make music and dance.

Alan Turing (British mathematician, 1936) was one of the most important visionary and theoreticians, proving that a universal calculator—known as the Turing machine—is possible. Turing’s central insight is that such a machine is capable of solving any problem as long as it may be represented and solved by an algorithm.

Newell and Simon (1955) designed “The Logic Theorist,” which is considered to be the first AI program that marks the development of modern AI.

John McCarthy in 1965 coined the term “artificial intelligence.”

AI IN ORAL MEDICINE AND RADIOLOGY

Dental radiography and explorer (or dental probe) are highly reliable diagnostic tools in detecting dental caries. Much of the screening and final diagnosis tend to rely on dentists’ experience.

The CNN algorithm to detect dental caries on periapical radiographs was developed by Lee et al.[3] The CNN algorithm to detect caries on intraoral images was proposed by Kühnisch et al.[4] The cost-effectiveness of AI for proximal caries detection with dentists’ diagnosis was compared by Schwendicke et al.[5] The results showed that AI was more effective and less costly.

AI IN ORTHODONTICS

Traditionally, orthodontists face challenges in diagnosing malocclusions due to the intricate cephalometric analysis involved. However, AI has revolutionized orthodontics by streamlining accurate diagnoses, optimizing personalized treatment plans, and predicting treatment outcomes with precision. AI simplifies the complex process, enhancing efficiency and reliability in orthodontic care. Recent studies have demonstrated the effectiveness of AI in determining the necessity of tooth extractions in orthodontics and radiographs to predict extraction needs with promising results.[6] Meanwhile, Jung and Kim achieved a remarkable 92% accuracy using an AI expert system for permanent tooth extraction decisions.[7]

AI has revolutionized orthodontic analysis, enhancing landmark identification and treatment planning. Studies have demonstrated AI’s impressive accuracy, including Park et al.’s[8] deep-learning algorithm precisely identifying cephalometric landmarks on radiographs, and Choi et al.’s[9] AI model achieving a 96% success rate in determining surgery/non-surgery cases using lateral cephalometric radiographs. Furthermore, Kok et al.’s[10] AI algorithms accurately determined growth and development stages with 77.02% accuracy using cervical vertebrae maturation on cephalometric radiographs, complementing traditional methods such as hand-wrist X-rays and cephalometric analysis. These advancements underscore AI’s potential to transform orthodontic diagnosis and treatment planning.

AI IN ORAL AND MAXILLOFACIAL PATHOLOGY

Oral cancer, the most severe form of Oral Mucosal and Facial Pathology (OMFP), poses a significant global health burden. According to the World Health Organization (WHO), oral cancer affects over 657,000 individuals annually, resulting in more than 330,000 fatalities worldwide. To combat this, researchers have explored AI for enhanced tumor and cancer detection, leveraging radiographic, microscopic, and ultrasonographic images.

CNN algorithms have proven effective in automatic cancer detection. Notably, Warin et al.[11] utilized CNN to identify oral potentially malignant disorders (OPMDs) and oral squamous cell carcinoma (OSCC) in intraoral optical images. Similarly, Abureville et al.[12] employed CNN to diagnose oral squamous cell carcinoma (SCC) from confocal laser endomicroscopy images, demonstrating its suitability for early SCC diagnosis.

AI IN ENDODONTICS

Periapical lesions detection

It might be difficult for clinicians to determine a diagnosis and a plan of treatment for teeth showing periapical lesions and/or their symptoms. IOPA and OPG are the two two-dimensional (2D) diagnostic methods that are most frequently employed in everyday clinical practice to detect apical periodontitis. Periapical lesions are often observed as radiolucency on radiographs. However, because the actual three-dimensional (3D) anatomy is condensed into a 2D image, the information gleaned from these periapical radiographs is unreliable.

The severity of periapical lesions with regard to the diagnosis of periapical pathology with the use of models was conducted by Mol et al.[13] and Carmody et al.[14] A deep learning algorithm model can detect periapical radiolucency on panoramic radiographs as accurately as 24 oral and maxillofacial surgeons according to the 142 out of 153 periapical lesions could be detected by the AI system, and this detection accuracy rate was 92.8% as found by Orhan et al.[15]

Additionally, Flores et al.[16] established a methodology to separate granuloma from periapical cysts using CBCT images. It is valued highly in clinical practice because it allows periapical granulomas to recover the following root canal therapy without the need for surgery.

Root fracture detection

A major outcome that may need root resection or tooth extraction, vertical root fractures (VRF) make up 2% to 5% of crown/root fractures. A clinician’s diagnostic options are usually limited by low sensitivity and clinical presentation of traditional radiography in the identification of vertical root fractures.

In a different research, periapical radiographs and CBCT images were used to create a neural network to identify VRFs in the teeth that were both intact and root-filled. In comparison to images from 2D radiographs, they found that fracture identification of roots on CBCT images was superior in relation to specificity, accuracy, and sensitivity.

Fractures were generated in second molars by Fukuda et al.[17] who used wavelets to analyze them using synthetic data. Despite a tiny sample size, steerable wavelets were successfully used to detect fractures in high-resolution CBCT images.

WORKING LENGTH DETERMINATION

Correct determination of working length (WL) is crucial for successful root canal treatment outcomes. One method used to assess WL is radiography. Other methods include digital tactile sense, electronic apex locators, the reaction of the patient to a paper point or file point placed into the root canal system, and CBCT imaging.

An artificial neural network (ANN) system was used in determining the working length and showed exceptional accuracy of 96%, which was higher than the accuracy compared to professional endodontists as studied by Saghiri et al.[18]

Morphology of root and root canal System

Understanding the different types of root and root canal systems is a crucial element in the effectiveness of nonsurgical root canal therapy. CBCT and periapical radiography have often been employed for this purpose. When compared to radiography, CBCT imaging has been shown to be more accurate in determining the root and root canal geometries.

In an evaluation of 433 cone-beam computed tomographic segmentations of teeth, the authors found that AI performed exactly as well as a human operator while working much faster by Lahoud et al.[19]

Retreatment predictions

According to the report of Campo et al.[20] for the prediction of the result of nonsurgical retreatment of the root canal with risks and benefits, a case-based reasoning paradigm was designed. In essence, the system advised on whether to retreat or not.

Prediction of the viability of stem cells

The neuro-fuzzy interference method of assessment of stem cells extracted from the tooth pulp in many regenerative treatments was used in a study by Bindal et al.[21] The scientists next evaluated the precision of the prediction provided by utilizing adaptive neuro-fuzzy interferences to forecast these stem cells’ survival following microbial invasion.

AI in periodontology

In periodontology, the Periodontal Screening Index (PSI) is commonly employed to measure clinical attachment loss; however, its reliability is limited. Currently, periodontal disease screening relies heavily on dental professionals’ expertise, which can lead to overlooked cases of localized periodontal tissue loss. To address this, AI has emerged as a diagnostic tool for periodontitis and classification of periodontal disease types. Notably, researchers have leveraged CNNs to enhance detection accuracy. For instance, Krois et al.[22] utilized a CNN to identify periodontal bone loss (PBL) on panoramic radiographs, whereas Lee et al.[23] assessed the effectiveness of a CNN algorithm in automatically detecting periodontally compromised teeth.

AI IN FORENSIC ODONTOLOGY

AI as a significant scientific advancement has been widely applied in forensic medicine, proving highly effective in determining biological age and gender in both healthy and ill individuals. It is also used to analyze bite marks and predict mandibular morphology. Dentists play a crucial role in identifying individuals in cases of child abuse, crime, sexual assault, mass disasters, and other legal matters. Their ethical responsibility drives them to seek justice for victims and their families, particularly when dental remains are the only available evidence. AI technology has been successfully applied in this field, delivering impressive results.

De Tobel et al.[24] successfully utilized an automated technique based on CNNs to estimate a person’s age by analyzing lower third molar development through panoramic radiographs.

Patil et al.[25] employed ANNs to determine gender using panoramic radiographs, yielding promising results.

Nino-Sandoval et al.[26] utilized an AI model based on ANNs to predict mandibular morphology, also demonstrating encouraging outcomes.

AI IN PROSTHODONTICS

A 3D-DCGAN network for crown generation, utilizing 3D data directly in the process, resulting in crown morphology closely resembling natural teeth was reported by Ding.[27] The integration of AI with CAD/CAM and 3D/4D printing creates a more efficient and streamlined workflow. AI has also been applied to tasks such as shade matching and predicting debonding in CAD/CAM restorations.

In addition to fixed prosthodontics, the design of removable prosthodontics presents greater challenges due to the need to consider a wider range of factors and variables. Current machine learning algorithms are increasingly focused on supporting the design of removable dentures, such as by classifying dental arch types and predicting facial appearance in edentulous patients.

USES OF AI IN DENTISTRY

• AI systems can assist clinicians in delivering high-quality dental care to their patients.

• Dentists can leverage AI as a supplementary tool to enhance the accuracy of diagnosis, treatment planning, and outcome prediction.

• Non-specialist dentists can benefit from diagnostic support through deep-learning systems.

• Automated systems can save significant time and boost clinicians’ efficiency, such as by automatically identifying and numbering teeth for electronic dental records.

• Using AI for second opinions can improve diagnostic accuracy.

• These systems also offer substantial value in forensic diagnosis.

CONCLUSION

New technologies are rapidly advancing and being adopted in the dental field, with AI emerging as one of the most promising innovations. When trained on unbiased data and properly developed, AI systems offer exceptional accuracy and efficiency. Studies show that AI-based automated systems perform at an outstanding level, often matching the precision and accuracy of trained dental specialists. In some cases, these systems have even surpassed human experts in terms of both performance and accuracy.

Machine intelligence has the potential to be humanity’s final invention, representing a monumental breakthrough in history. However, it could also bring significant risks if we do not learn how to manage it responsibly.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.