Artificial Intelligence in Dentistry

 Artificial Intelligence is a constantly evolving reality.

Image created by artificial intelligence.

The advent of digitization in the dental sector has meant a paradigm shift for professionals dedicated to this specialty. However, our concept of "new technologies" may soon take on a new way of being seen and understood. In addition, the increasingly comprehensive introduction of Artificial Intelligence (AI) resources in medicine have also precipitated a rapid advancement of AI in dentistry.

What is Artificial Intelligence (AI)?

The term Artificial Intelligence (AI) was coined by the computer scientist John McCarthy in 1956, but it was Frank Rosenblat 1957 who designed the first neural network or "intelligent brain."

AI can be defined as the ability of specific machines to mimic human knowledge and behavior by implementing a sequence of algorithms (artificial neurons). It is an automated way of performing activities related to human abilities, including the ability to learn, make decisions, and solve problems. This concept aims to facilitate and improve professionals' diagnostic and therapeutic capacity in the bio-health world by feeding the "artificial brain" through infinite data sources.

When we speak of "Big Data" in Artificial Intelligence, we refer to the set of data or combinations of data that make it possible to establish a reference point from all the information provided to the system so far. It is, therefore, the engine from which AI is nourished. Thanks to this, processing a large volume of data is speeded up.

This type of revolutionary technology (AI) is composed of a sequence of "artificial neurons" whose connection creates different layers that, together with the application of specific mathematical operations, develop a network that allows the resolution of a particular task, such as the classification of structures in radiographic images.

AI encompasses the following fields of study:

1. Machine Learning (ML): these are systems of algorithms with the ability to learn from the "data" that is provided without being explicitly programmed to do so. It is mainly designed to predict results based on previously obtained data (from which it has been nurtured and, therefore, learned).

2. Deep Learning (DL): this is a subtype of ML in which artificial neural networks can adapt and learn from large amounts of data. They are usually responsible for automatically identifying patterns and/or common characteristics in the supplied information (data).

Despite the complexity of these mathematical systems, we must bear in mind that we have been coexisting with them for years. A clear example we may all be familiar with is the automatic unlocking of our mobile devices through facial or fingerprint recognition.

Automatic grammar correction systems present when we write messages or documents, or even what we visualize on social media platforms, are based on AI.

Applications of AI in Medicine:

So far, the use of AI in different medical specialties has improved in diagnosing patients' and professionals' decision-making processes in complex and challenging cases.

Among all medical specialties, very positive results have been described for its use in detecting various cardiovascular, retinal-ocular, and dermatological pathologies. In fact, the latter is probably the area in which the most significant advances have been made about this new technology: studies have been published on the detection of nevi and their classification (malignant or benign), as well as on the determination of the severity of common inflammatory diseases (such as psoriasis, urticaria or some bullous autoimmune pathologies). Recently, the well-known company Google (Google Health) launched an AI-based application capable of detecting common skin conditions with the exact prediction of a dermatologist. Its results were published in the prestigious journal Nature Medicine.

AI applications in dentistry:

But what applications can we find today in our sector? As unlikely as it may seem, we have been coexisting with AI for a few years. For example, some leading dental brands have developed smart toothbrushes that evaluate the patient's hygiene technique, telling them where to deepen their dental hygiene routine.

Another clear example of AI is the automation of orthodontic treatment plans developed by multiple companies through 3D scans of our patient's mouths. This case has been the most economical and labor-relevant application in recent years.

However, it is essential to know the current status of these technologies and how they may affect us in the future. So far, several scientific studies have described how this type of "new technology" is used in the prosthetic field to help the clinician in the precision of tooth preparation for fixed prostheses or in the registration of mandibular movements. In addition, there are specific systems capable of predicting, in the periodical check-ups of our patients, the possibility of crown debonding or loosening of crowns on implants. Also important is the ability of specific intraoral scanners to recognize and eliminate unimportant adjacent structures (tongue, cheeks, etc.) to design and create prosthetic restorations.

However, what is the top line of dental research at the moment? Diagnostics. Currently, studies have been described in which the diagnostic capacity of different types of neural networks has been tested in diagnosing photographic images (detection of bacterial plaque, caries, etc.) and radiographic images (2D and 3D).

In recent years, much weight has been given to how these types of applications could improve the diagnostic capacity of dental professionals, especially considering that not everyone is equally trained or has the same clinical experience (which, undoubtedly, can interfere with their critical and analytical capacity). This is why AI can be beneficial: its automation can guide the clinician in making better decisions to ensure a good patient prognosis.

In our research group, we have started to perform the first evaluation studies of 2D AI-based radiological detection programs available on the Web, in which we evaluated the capability of an AI-based program (Denti.Ai) for the detection of carious pathologies in bitewing radiographs and for the determination of an overall diagnosis of the patient including which dental structures and treatments were or were not present in panoramic radiographs. However, as far as 2D imaging is concerned, the main focus of attention is currently focused on the detection of pathologies (such as mandibular tumors, root fractures, or possible radiological signs associated with Sjögren's Syndrome) and on the identification and classification of various types of dental implants (brand name).

In the last year, studies have been increasing that include 3D images (Diagnocat). Still, their progress could be faster due to the complexity of the processing of the radiological tests used. Their main objective is the segmentation and three-dimensional classification of the different oro-facial structures. Nevertheless, there are articles published in important international journals describing its use for the detection of periapical lesions, dental canals, etc.

In short, AI is beginning to enter all dental specialties without us being fully aware. It is a future that is already in our present. It is only a matter of time before it fully enters our fields of work.

Source of information: Artificial intelligence in healthcare