|Year : 2020 | Volume
| Issue : 1 | Page : 27-31
Artificial intelligence: In modern dentistry
V Bindushree, RJ Sameen, Vijeev Vasudevan, TG Shrihari, D Devaraju, Nimi Susan Mathew
Department of Oral Medicine and Radiology, Krishnadevaraya College of Dental Sciences and Hospital, Bengaluru, Karnataka, India
|Date of Submission||01-Aug-2020|
|Date of Decision||20-Feb-2020|
|Date of Acceptance||02-Apr-2020|
|Date of Web Publication||28-Mar-2020|
Department of Oral Medicine and Radiology Krishnadevaraya College of Dental Sciences Bangalore, Karnataka
Source of Support: None, Conflict of Interest: None
Artificial intelligence (AI) is the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings. There is marked increase in evolution of AI from the last decade, which has been showing tremendous improvement and dentistry is no exception. AI has its importance in dentistry, especially in oral medicine and radiology, which includes patient diagnosis, storage of patient data, and the assessment of genetic information which will provide improved healthcare for patients. Regardless of many improvements and advances, AI is still in its teething stage, but its potential is boundless. This article reviews on how this technology is tremendously utilized for easy and early diagnosis, proper treatment of lesions of oral cavity, advanced breakthroughs in image recognition techniques, screening of suspicious premalignant, and malignant changes of oral cavity with satisfying outcome. A thorough knowledge regarding the adaptation of technology will not only help in better and precise patient care but also reducing the work burden of the clinician.
Keywords: Artificial intelligence, machine learning, modern dentistry, neural networks
|How to cite this article:|
Bindushree V, Sameen R J, Vasudevan V, Shrihari T G, Devaraju D, Mathew NS. Artificial intelligence: In modern dentistry. J Dent Res Rev 2020;7:27-31
|How to cite this URL:|
Bindushree V, Sameen R J, Vasudevan V, Shrihari T G, Devaraju D, Mathew NS. Artificial intelligence: In modern dentistry. J Dent Res Rev [serial online] 2020 [cited 2021 Jan 27];7:27-31. Available from: https://www.jdrr.org/text.asp?2020/7/1/27/281504
| Introduction|| |
The concept of artificial intelligence (AI) was given by John McCarthy in the year 1956. Constant search for the model has led to the development of AI. The definition of AI says that it is a field of science and engineering concerned with the computational understanding of what is commonly called intelligent behavior and with creation of artifacts that exhibit such kind of behavior. AI project intelligent systems are evolving in their own environment to maximize their own chances of success.
So what is AI? AI is sometimes addressed as intelligence of machines, i.e., work performed by machine, in difference with the natural intelligence performed by humans and other different animals. In other words, AI may be referred as a subject which deals with computerized models that can think and act or perform tasks rationally., AI systems can help nonspecialists to gain information with more accuracy and at the level of experts. Artificial neural networks (ANNs) are inspired by human neural system or biological nervous system, which are highly interconnected networks of computer processors systems. ANNs simulate the neural signal transmission and the human brains serves as an essential part of AI., Programming languages are the principal tools of AI in understanding these symbolic information systems.
The advantage of these ANNs is that, it can favor the dental professionals worldwide to connect with each other. Currently, the use of AI is proliferating a way forward to text-based, image-based dental practice. This article aims to provide foresight into the current concepts and the likely future prospects of AI in dentistry.
Artificial neural networks
Development of ANNs is based on the human brain structure or biological brain structure and they can recognize pattern such as human brain, manage data, and learning. The most important advantage of ANNs is that this system solves problems which are too complex to conventional techniques and also, those that do not have an algorithmic solution can be solved with the help of ANN. They are utilized in various fields of medicine like, for diagnosis or diagnostic systems, biomedical analogies, development of drugs, and image analysis.
In 1957, Frank Rosenblatt invented the perception algorithm which was designed for recognition of the images. It had the “neurons” which is randomly connected with an array of 400 photocells. The signals are sent by the neurons only when the aggregate signals crosses the threshold of the specific neuron. Usually, the neurons are settled or aggregated in the form of layers [Figure 1]. Each layer will perform differently or gives different signals depending on their inputs. We can get different transformations from each layer. Signals travel from the input layer (the first layer), to the output layer (the last layer), mostly after passing the layers several times [Figure 2]. Potentiometers are used to encode the weights and electronic motors will perform weight updates during learning., Back propagation learning was proposed by Paul Webros in 1974. It is a method used in AI to calculate a gradient that is needed in the calculation of the weights to be used in the networks. This is used by health professionals to diagnose the disease early and to communicate with fellow professionals worldwide; thus, more effective treatment is possible to provide for the patient.
|Figure 1: An artificial neural network is an interconnected group of nodes, inspired by a simplification of neurons in a brain. Here, each circular node represents an artificial neuron and an arrow represents a connection from the output of one artificial neuron to the input of another|
Click here to view
|Figure 2: Neuron and myelinated axon, with signal flow from inputs at dendrites to outputs at axon terminals|
Click here to view
Now, there are two sub-concepts which will divide the whole range of meanings which are presently derived by the term “AI.” The coexistence of the concepts of strong and weak AI can be seen as a result of the recognition of the limits of mathematical and engineering concepts that dominated definitions of AI in the first place.
| Weak Artificial Intelligence – Definition and Its Concept|| |
The concept of weak AI intends to improve the cognitive behavior and judgmental capacity of computer system to make them inherent in those computing systems, denying the unreasonable reduction, and in attempt to reproduce the human intelligence, which is predicted and planned by strong intelligence., The meaning of weak AI is, the system in which human beings can gain advantage of few medical and logical mechanisms in which the intellectual activities to execute efficiently by intelligence neural networks like human can perform.
| Strong Artificial Intelligence – Definition and Its Concept|| |
Strong AI means a system that works in the same way as human intelligence through unnatural, software construction, and artificial hardware. It is a theoretical form of machine intelligence., The key feature of strong AI involves reasoning ability, puzzle-solving, judgment making, planning, learning, and communicating, i.e., the capacity of machines to perform human tasks and replicate or reproduce human behavior efficiently.
It is one of the sections in AI that provides knowledge to computer systems through data and observations without actually being programmed. This allows a computer to correctly generalize a setting by tuning or adjusting the parameters within the algorithm to achieve the fitness between the input (i.e. text, image, or video data fed into the algorithm) and output (i.e. classification). For example, for a machine learning (ML) algorithm can recognize or detect a lymph node in the head-and-neck image as normal or abnormal provided it is trained radiologist by analyzing thousands of such images which are labeled as normal or abnormal., To sum it up ML algorithms are trained to provide a correct specific answer by examining or learning a huge number of procedural tests that have been hand-labeled.,
Is a subtype of ML in which the computer algorithm systems studies about the features necessary to categorize the data that are provided. This does not require a hand labeled data like ML [Figure 3].
Is a subgroup of representation learning relying on multiple processing layers (hence, deep) to seek knowledge about representations of data with one or more multiple layers of abstraction. This algorithm uses multiple layers to detect simple features such as line, edge and texture to complex shapes, lesions, or whole organs in a hierarchical structure., Basis of any radiologic interpretation is the logical elimination of possible diagnosis. Hence, deep learning can be exceptionally good by learning a specific type of hierarchical normal representation of particular image from a huge number of normal examinations.,
Artificial intelligence in different fields of dentistry
Artificial intelligence in patient management
Virtual dental assistants which are based on AI can perform various functions and tasks with greater accuracy in the dental clinic, minimal errors, and less workforce compared to humans. In the departments such as oral medicine and radiology, oral pathology, it can be used to arrange appointments, managing insurance and article works as well as helping in diagnosis or planning treatment. It is very helpful in notifying the dentist regarding patients' complete medical and dental history as well as other oral hygiene habits, food and diet habits and habits such as alcoholism and smoking. In dental emergencies, the patient has an option of emergency tele-assistance, especially when the practitioner is unavailable. Thus, a comprehensive virtual data of the patients can be generated which will help in providing ideal treatment for the patient in long run.
Artificial intelligence in diagnosis and treatment
For a successful clinical practice, correct diagnosis is a strong foundation. In this concern, efficiently trained neural networks can be a precious present to diagnosticians, particularly in the diseases and conditions with multifactorial cause or etiology. To say, recurrent aphthous ulceration is one of the conditions without a specific etiology or with multiple etiology, where the clinical diagnosis is given based only on the recurrence of the lesion and excluding the other factors. In this regard, AI can be considered as one of the useful and ideal modalities in diagnosing and treatment planning or treating of oral mucosal lesions and can be utilized in examining and grouping doubtful or unsure altered mucosa which is considered to show premalignant and malignant changes. Even minimal to minute changes at the level of single pixel which might go undetectable by the human eye can be detected. AI can precisely prognosticate the predisposition of genes in oral cancer for a large population.
Artificial intelligence in oral and maxillofacial surgery
The huge application of AI in this field is the development of robotic surgery where the stimulation of human body motion and human intelligence is shown by AI. Successful clinical application in image-guided surgery in the cranial area includes oral implant surgery, removal of tumor and foreign bodies, biopsy, and temporomandibular joint surgery.
Few comparative studies in the literature of using AI in oral implant surgery indicate significantly more accuracy compared to manual freehand procedures even if performed by experienced surgeons. In addition, no significant difference between experienced surgeon and trainees were identified. In spite of that shorter operation time, safer manipulation around delicate structures and higher intraoperative accuracy has been recognized with the help of AI. Image guidance allows thorough surgical resection which may decrease the requirement of revision procedures.
Artificial intelligence in prosthetic dentistry
In order to render ideal flawless esthetic prosthesis for the patient various factors such as facial measurements, anthropological calculations, ethnicity, and patient preferences has been integrated by a design assistant which uses AI (RaPid) for use in prosthodontics. RaPiD integrates computer-aided design (CAD), knowledge-based systems and databases, recruit a logic-based information as a unifying medium. AI in the form of CAD/computer-aided manufacturing (CAM) application which is used in dentistry, in which the dental restorations are processed and finish restorations through fine grinding process of ready ceramic blocks. It is used in inlay manufacturing, onlay manufacturing, as well as crowns and bridges. CAD/CAM technique essentially helps to create two-dimensional and three-dimensional models and numerically controlled mechanics aids in their materialization process. It has replaced the time-consuming procedures like laboratory procedures of routine casting and reducing the human mistakes in final prosthesis.
Artificial intelligence in orthodontics
Diagnosis and treatment planning can be done in orthodontics by the analysis of radiographs and photographs by intraoral scanners and cameras which works on the principles of AI. This eliminates the requirement for making patient impressions as well as several laboratory steps that are usually followed. From this, the results are usually much more accurate compared to human perception. The tooth movement and final treatment outcome can be predicted by using algorithms and statistical analysis.
Applications of artificial intelligence
Adapting of AI in maxillofacial radiology, its clinical applications can be divided into three types [Figure 4].
|Figure 4: Table representing clinical applications of artificial intelligence|
Click here to view
- Clinical workflow
- Types of applications
- Classes of use cases.
Advantages of artificial intelligence
- It is an effective tool or aid to recognize patterns, predict events, and for grouping objects
- Radiology departmental workflow improvement through identify patients most at risk of missing appointments, precision scheduling, and empower individually tailored examination protocols
- ML directly with medical data can help in preventing errors due to cognitive bias or human-induced bias.
• Requires a very huge and sound powerful database of knowledge, if not may result in irrelevant answers when given with images outside of their set of knowledge. For example, when image techniques are not appropriate or if there are any artifacts may result in the faulty interpretation of image
- May not adapt with new imaging software or new machine immediately
- Not all the algorithms used are apt for clinical application. More trials to recommend apt analytic programs for different scenarios.
Future recommendations in the field of artificial intelligence
- AI terminology and hierarchy should be familiar to the radiologists
- Radiology programs should begin to incorporate health informatics, computer science, and statistics courses in their curriculum
- Training should be given to the radiologist for logic, statistics, and data science and they should know about other information related to genomics and biometrics
- For supervised learning, radiologists should know about the challenges regarding the preparation of training datasets.
| Conclusion|| |
The usage of AI is growing rapidly in everyday life. Dentists are not lagging in implementing technology. Thus, learning and understanding the various concepts and the techniques involved will have a clear advantage in upcoming days, which has more scope for the developing technology. There is tremendous potential for research in AI in medicine and dentistry. The research should be integrated with clinical practice for better results. Even though advanced sign natural language processing, image recognition, neural networking and speech recognition are on the anvil the high initial costs can often be a deterrent. AI can certainly be tool in making significant progress in delivering better healthcare to the patient, but in no way can replace human knowledge, skills, and power of judgment.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Alexander B, John S. Artificial intelligence in dentistry: Current concepts and a peep into the future. Int J Adv Res 2018;6:1105-8.
Deshmukh SV. Artificial intelligence in dentistry. J Int Clin Dent Res Organ 2018;10:47-8. [Full text]
Park WJ, Park JB. History and application of artificial neural networks in dentistry. Eur J Dent 2018;12:594-601.
] [Full text]
Prasad S, Pai A, Yaji A,et al
. Artificial intelligence in dento-maxillofacial radiology. Acta Sci Dent Sci 2019;3:116-21.
Kalappanavar A, Sneha S, Annigeri RG. Artificial intelligence: A dentist's perspective. J Med Radiol Pathol Surg 2018;5:2-4.
Talari AC, Rehman S, Rehman IU. Advancing cancer diagnostics with artificial intelligence and spectroscopy: Identifying chemical changes associated with breast cancer. Expert Rev Mol Diagn 2019;19:929-40.
Vashisht A, Choudhary E. Artificial intelligence; mutating dentistry 2019. IJRAR 2019;6: (E-ISSN 2348-1269, P- ISSN 2349-5138).
Widmann G. Image-guided surgery and medical robotics in the cranial area. Biomed Imaging Interv J 2007;3:e11.
Kattadiyil MT, Mursic Z, AlRumaih H, Goodacre CJ. Intraoral scanning of hard and soft tissues for partial removable dental prosthesis fabrication. J Prosthet Dent 2014;112:444-8.
Majumdar B, Sarode SC, Sarode GS, Patil S. Technology: Artificial intelligence. Br Dent J 2018;224:916.
Tunjugsari V, Sabiq A, Sofro AS, Kardiana A. Investigating CDSS success factors with usability testing. Int J Adv Comput Sci Appl (IJA CSA) 2017;8:548-54.
Kwon HB, Park YS, Han JS. Augmented reality in dentistry: A current perspective. Acta Odontol Scand 2018;76:497-503.
Khanna SS, Dhaimade PA. Artificial intelligence: Transforming dentistry today. Indian J Basic Appl Med Res 2017;6:161-7.
Hwang JJ, Sergei A, Efros AA, Yu SX. Learning Beyond Human Expertise with Generative Models for Dental Restoration. CoRR abs/1804.00064; 2018.
Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394-424.
International Agency for Research on Cancer; 2018. Latest Global Cancer Data: Cancer Burden Rises to 18.1 Million New Cases and 9.6 Million Cancer Deaths in 2018. Available from: http://gicr.iarc.fr/
. [Last accessed on 2019 Sep 12].
Surmacki JM, Woodhams BJ, Haslehurst A, Ponder BA, Bohndiek SE. Raman micro-spectroscopy for accurate identification of primary human bronchial epithelial cells. Sci Rep 2018;8:12604.
Tree AC, Huddart R, Choudhury A. Magnetic resonance-guided radiotherapy – Can we justify more expensive technology? Clin Oncol (R Coll Radiol) 2018;30:677-9.
Chen AM, Chin R, Beron P, Yoshizaki T, Mikaeilian AG, Cao M. Inadequate target volume delineation and local-regional recurrence after intensity-modulated radiotherapy for human papillomavirus-positive oropharynx cancer. Radio Ther Oncol 2017;123:412-8.
Weinbaum D, Veitas V. Open ended intelligence: The individuation of intelligent agents. J ExpTheor Artif Intell 2017;29:371-96.
Zhu J, Wen D, Yu Y, Meudt HM, Nakhleh L. Bayesian inference of phylogenetic networks from bi-allelic genetic markers. PLoS Comput Biol 2018;14:e1005932.
Tang A, Tam R, Cadrin-Chênevert A, Guest W, Chong J, Barfett J, et al
. Canadian Association of radiologists white paper on artificial intelligence in radiology. Can Assoc Radiol J 2018;69:120-35.
Drozdzal M, Chartrand G, Vorontsov E, Shakeri M, Di Jorio L, Tang A, et al
. Learning normalized inputs for iterative estimation in medical image segmentation. Med Image Anal 2018;44:1-3.
Rajpurkar P, Irvin J, Zhu K, Yang B, Mehta H, Duan T, et al
. CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning (Submitted on 14 Nov 2017 (v1), last revised 25 Dec 2017 (this version, v3)). https://arxiv.org/abs/1711.05225
Alkasab TK, Bizzo BC, Berland LL, Nair S, Pandharipande PV, Harvey HB. Creation of an open framework for point-of-care computer-assisted reporting and decision support tools for radiologists. J Am Coll Radiol 2017;14:1184-9.
Gillies RJ, Kinahan PE, Hricak H. Radiomics: Images are more than pictures, they are data. Radiology 2016;278:563-77.
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. doi:10.1259/dmfr.20160107.
Miller DD, Brown EW. Artificial intelligence in medical practice: The question to the answer? Am J Med 2018;131:129-33.
Aktolun, C. Artificial intelligence and radiomics in nuclear medicine: potentials and challenges. Eur J Nucl Med Mol Imaging 2019;46:2731-36. Doi: doi.org/10.1007/s00259-019-04593-0.
Jiawei Su, Danilo Vasconcellos Vargas, Sakurai KouichiSu J, et al
. One Pixel Attack for Fooling Deep Neural Networks. arXiv Preprint arXiv: 1710.08864; 2017.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]