We are on the cusp of a dental technology revolution. Advances in artificial intelligence (AI) and machine learning have the potential to improve oral health for everyone. However, if we want to see these technologies reach their full potential, we need to address data quality issues first. Tackling challenges with proper data collection and annotation are a stepping stone toward revolutionizing the profession in more ways than we know. By improving said data collection and annotation processes, we can enable AI to do amazing things for dentistry and improve the lives of all people who rely on it. 

The Current Role of AI in Dentistry

There is no doubt that artificial intelligence (AI) is transforming dentistry. From diagnosis and treatment planning to dental lab workflows, AI is making dentistry more efficient and accurate. For example, AI-based smile design tools can now create a 3D model of a patient’s teeth that can be used to plan orthodontic treatment or tooth whitening. In addition, AI is being used to develop more effective and customized dental prosthetics. With AI, dentists can now create highly realistic dentures, crowns, and bridges that better match a patient’s natural teeth. Dentists can also use AI to detect cavities and other problems at an early stage, which can help prevent more serious issues down the road. As dentistry continues to evolve, it is clear that AI will play an increasingly important role.

The Obstacle to Progress

There is one major obstacle that AI must overcome before it can truly revolutionize dentistry: the lack of quality data. In order for AI to be effective, it needs large amounts of high-quality data to learn from. Unfortunately, this is something that dentistry lacks. The vast majority of dental data is unstructured, unorganized, and unstandardized, making it difficult for AI algorithms to learn from. This problem is compounded by the fact that dental data is spread across a wide variety of silos and is protected by endless internal privacy policies, making it even harder to access and use. And, let’s not forget that much of the dental data that exists today is manually recorded, sometimes even by hand, which makes it hard to understand and digitize.

If we want AI to truly transform dentistry, we need to fix our data problem. We need to find a way to collect and store dental data in a way that makes it easy for AI algorithms to learn from. We also need to find a way to share this data with other dentists and scientists so that they can benefit from it as well. Only then will we be able to realize the full potential of AI in dentistry.

The Fix

As we have established, dentistry has long been a data-driven profession, relying on dentists to manually record and document patient information. However, the quality of dental data has long been an issue, due in part to the lack of standardization in dentistry. This has led to a number of problems, including duplicate records, incorrect or missing data, and other errors.

Ironically, the most effective way to help AI revolutionize dentistry is to use even more AI to improve the quality of dental data, which can be used to automatically record and document patient information, eliminating the need for manual data entry. AI can also help to identify errors in dental data and correct them automatically. In addition, AI can be used to standardize and encrypt dental data, making it easier to share and compare records between dentists. As a result, AI holds great promise for improving the quality of dental data. That is if we use it to its full potential.

The Future of AI in Dentistry

Unlike many other fields, dentistry is one of the few professions that has not been deeply impacted by artificial intelligence (AI). However, this may soon change — if we improve the data collection and annotation obstacles, of course. AI is becoming increasingly good at analyzing data, and dentists generate a lot of data. Every time a dentist sees a patient, they take X-rays, chart the teeth, and make notes on the treatment. This data can be used to train AI algorithms to spot patterns and predict outcomes. In the future, AI may be used to diagnose dental problems, plan treatments, and even carry out procedures. Of course, dentists will still be needed to supervise AI-assisted dentistry and provide care for patients. But as AI gets better at handling routine tasks, dentists will have more time to focus on complex cases and research. Ultimately, AI has the potential to transform dentistry for the better, making it more efficient and effective. But, as we have seen, this will only be possible if we fix our data problem first.

Dr. Laura Nicolas
Dr. Alberto Teramoto

Department of Orthodontics. Universidad Latinoamericana Mexico City, Mexico.

Introduction

There is no doubt that resolving impacted teeth is very difficult with an inadequate treatment plan which lacks all necessary information. Moreover, it’s vital to know where impacted teeth are located and more importantly, what their relationship is to other teeth, roots and even the mandibular nerve.

After the third molars, maxillary canines have the highest frequency of impacted localization, with a prevalence ranging from 1% to 3% (1-4), and it is more frequent in female patients than male with a ratio of 2:1 (5). If surgical treatment is necessary, it requires a great deal of time and is a challenge to treat without the right information. This is why the accurate location of impacted maxillary canines is so crucial when deciding whether to treat orthodontically or with surgical intervention.

During our daily practice, the first radiographic image usually taken to support the clinical examination is the panoramic radiograph, and sometimes in order to have a better localization of impacted teeth, we can improve diagnosis by combining two or more bi-dimensional images such as: occlusal and periapical, which allow the localization of impacted canines, treatment planning, and evaluation of the treatment result.

Importance of CBCT

It is well known that the accuracy of these two-dimensional radiographic techniques is severely limited, increasing the risk of error. By adding a third dimension to the radiographic information, clinicians can improve diagnostic accuracy and treatment efficacy – using Cone Beam Computed Tomography (CBCT), clinicians can take advantage of 3D information provided by a low radiation dose, and at relatively low costs. CBCT provides information that is not revealed through traditional radiographic analysis, and is therefore recommended in cases of impacted teeth or craniofacial structural irregularities.

Although we have all these 3D images sometimes they are not enough, CBCT data offers much more than what we are able to use for imaging. DICOM data can also be used for segmentation, a process that separates structures of interest from the background and from each other, and for which a number of algorithms have been developed in image processing. Segmentation of an object is achieved either by identifying all pixels or voxels that belong to the object, or by locating those that form its boundary. Segmenting teeth is not an easy task. The following problems may arise when segmenting teeth from CBCT images:

1) DICOM data is acquired in upper-lower jaw in occlusion, so it is difficult to distinguish a tooth from its neighboring teeth along their occlusal surface due to the lack of gray-scale value changes.

2) It is also hard to separate a tooth from alveolar bone by similar densities.

3) Often, teeth possess similar shapes, making it difficult to identify different tooth types.

While there are many open source tools available for performing segmentation procedures, obtaining good quality segmentation takes time and requires a long learning curve.

Using artificial intelligence technology, Orca Dental AI created a unique system in a 3D controllable STL format that includes not only teeth segmentation but also cephalometric and airway volume calculations, which make this process more comprehensive, easy to use and completely informative in disclosing all the impacted relationships.

To aid in an improved diagnostic approach and improve treatment outcomes, Data Imaging and Communication in Medicine (DICOM) files of the CBCT can also be converted into a 3D model using a 3D printer (Zenith D DENTIS Co. LTD). Therefore, this tool is very useful to visualize and be able to feel 3D conditions of teeth that are on the verge of falling out, since this process offers an extra diagnostic dimension. When analyzing this data, we can decide which is the best treatment plan, and for example decide if an orthodontic treatment to align an ectopic canine is better than a surgical procedure because ectopic canines are hard to align.

CASES

Case 1

Since the root of the impacted canine is located close to the vestibular side, when the first premolar is extracted, the canine can easily be moved into its ideal position. (Fig.1 a-h)

 

Case 2

Due to the horizontal position of the right impacted canine, surgical options are clearly the best solution. (Fig.2 a-f).

Conclusions

The use of CBCT in clinical practice can be for a variety of clinical problems, including impacted teeth, craniofacial anomalies, TMJ analysis, and analysis of the upper airway.

Combining CBCT images with 3D segmentations impressions and videos enables a better diagnosis and helps to decide whether to treat these cases orthodontically or surgically.

Special thanks

Orca Dental AI for providing segmentation, video analysis, and cephalometric analysis.

Ideas Dentales Mexico for printing  3D models.

References

1. Mason C, Papadakou P, Roberts GJ. The radiographic localization of impacted maxillary canines: a comparison of methods. Eur J Orthod 2001;23:25-34.
2. Preda L, La Fianza A, Di Maggio EM, Dore R, Schifino MR, Campani R, et al. The use of spiral computed tomography in the localization of impacted maxillary canines. Dentomaxillofacial Radiol 1997;26:236-41.
3. Stewart JA, Heo G, Glover KE, Williamson PC, Lam EW, Major PW. Factors that relate to treatment duration for patients with palatally impacted maxillary canines. Am J Orthod Dentofac Orthop 2001;119:216-25.
4. Walker L, Enciso R, Mah J. Three-dimansional localization of maxillary canines with cone-beam computed tomography. Am J Orthod Dentofac Orthop 2005;128:418-23.
5. Peck S, Peck L, Kataja M. The palatally displaced canine as a dental anomaly of genetic origin. Angle Orthod. 1994;64:249-56.
6. Ericson S, Kurol J. CT diagnosis of ectopically erupting maxillary canines – a case repost. European Juornal fo Orthodontics 1988;10:115-21.