[The challenges of artificial intelligence in odontology]

Artificial intelligence has numerous potential applications in dentistry, as these algorithms aim to improve the efficiency and safety of several clinical situations. While the first commercial solutions are being proposed, most of these algorithms have not been sufficiently validated for clinical use. This article describes the challenges surrounding the development of these new tools, to help clinicians to keep a critical eye on this technology.

Automatic 3-Dimensional Cephalometric Landmarking via Deep Learning

The increasing use of 3-dimensional (3D) imaging by orthodontists and maxillofacial surgeons to assess complex dentofacial deformities and plan orthognathic surgeries implies a critical need for 3D cephalometric analysis. Although promising methods were suggested to localize 3D landmarks automatically, concerns about robustness and generalizability restrain their clinical use. Consequently, highly trained operators remain needed to perform manual landmarking. In this retrospective diagnostic study, we aimed to train and evaluate a deep learning (DL) pipeline based on SpatialConfiguration-Net for automatic localization of 3D cephalometric landmarks on computed tomography (CT) scans. A retrospective sample of consecutive presurgical CT scans was randomly distributed between a training/validation set (n = 160) and a test set (n = 38). The reference data consisted of 33 landmarks, manually localized once by 1 operator(n = 178) or twice by 3 operators (n = 20, test set only). After inference on the test set, 1 CT scan showed "very low" confidence level predictions; we excluded it from the overall analysis but still assessed and discussed the corresponding results. The model performance was evaluated by comparing the predictions with the reference data; the outcome set included localization accuracy, cephalometric measurements, and comparison to manual landmarking reproducibility. On the hold-out test set, the mean localization error was 1.0 ± 1.3 mm, while success detection rates for 2.0, 2.5, and 3.0 mm were 90.4%, 93.6%, and 95.4%, respectively. Mean errors were -0.3 ± 1.3° and -0.1 ± 0.7 mm for angular and linear measurements, respectively. When compared to manual reproducibility, the measurements were within the Bland-Altman 95% limits of agreement for 91.9% and 71.8% of skeletal and dentoalveolar variables, respectively. To conclude, while our DL method still requires improvement, it provided highly accurate 3D landmark localization on a challenging test set, with a reliability for skeletal evaluation on par with what clinicians obtain.

Fully automatic segmentation of craniomaxillofacial CT scans for computer-assisted orthognathic surgery planning using the nnU-Net framework

OBJECTIVES

To evaluate the performance of the nnU-Net open-source deep learning framework for automatic multi-task segmentation of craniomaxillofacial (CMF) structures in CT scans obtained for computer-assisted orthognathic surgery.

METHODS

Four hundred and fifty-three consecutive patients having undergone high-resolution CT scans before orthognathic surgery were randomly distributed among a training/validation cohort (n = 300) and a testing cohort (n = 153). The ground truth segmentations were generated by 2 operators following an industry-certified procedure for use in computer-assisted surgical planning and personalized implant manufacturing. Model performance was assessed by comparing model predictions with ground truth segmentations. Examination of 45 CT scans by an industry expert provided additional evaluation. The model's generalizability was tested on a publicly available dataset of 10 CT scans with ground truth segmentation of the mandible.

RESULTS

In the test cohort, mean volumetric Dice similarity coefficient (vDSC) and surface Dice similarity coefficient at 1 mm (sDSC) were 0.96 and 0.97 for the upper skull, 0.94 and 0.98 for the mandible, 0.95 and 0.99 for the upper teeth, 0.94 and 0.99 for the lower teeth, and 0.82 and 0.98 for the mandibular canal. Industry expert segmentation approval rates were 93% for the mandible, 89% for the mandibular canal, 82% for the upper skull, 69% for the upper teeth, and 58% for the lower teeth.

CONCLUSION

While additional efforts are required for the segmentation of dental apices, our results demonstrated the model's reliability in terms of fully automatic segmentation of preoperative orthognathic CT scans.

KEY POINTS

• The nnU-Net deep learning framework can be trained out-of-the-box to provide robust fully automatic multi-task segmentation of CT scans performed for computer-assisted orthognathic surgery planning. • The clinical viability of the trained nnU-Net model is shown on a challenging test dataset of 153 CT scans randomly selected from clinical practice, showing metallic artifacts and diverse anatomical deformities. • Commonly used biomedical segmentation evaluation metrics (volumetric and surface Dice similarity coefficient) do not always match industry expert evaluation in the case of more demanding clinical applications.

A new protocol to accurately track long-term orthodontic tooth movement and support patient-specific numerical modeling

Numerical simulation of long-term orthodontic tooth movement based on Finite Element Analysis (FEA) could help clinicians to plan more efficient and mechanically sound treatments. However, most of FEA studies assume idealized loading conditions and lack experimental calibration or validation. The goal of this paper is to propose a novel clinical protocol to accurately track orthodontic tooth displacement in three-dimensions (3D) and provide 3D models that may support FEA. Our protocol uses an initial cone beam computed tomography (CBCT) scan and several intra-oral scans (IOS) to generate 3D models of the maxillary bone and teeth ready for use in FEA. The protocol was applied to monitor the canine retraction of a patient during seven months. A second CBCT scan was performed at the end of the study for validation purposes. In order to ease FEA, a frictionless and statically determinate lingual device for maxillary canine retraction was designed. Numerical simulations were set up using the 3D models provided by our protocol to show the relevance of our proposal. Comparison of numerical and clinical results highlights the suitability of this protocol to support patient-specific FEA.

Biplanar Low-Dose Radiograph Is Suitable for Cephalometric Analysis in Patients Requiring 3D Evaluation of the Whole Skeleton

BACKGROUND

The biplanar 2D/3D X-ray technology (BPXR) is a 2D/3D imaging system allowing simultaneous stereo-corresponding posteroanterior (PA) and lateral 2D views of the whole body. The aim of our study was to assess the feasibility of cephalometric analysis based on the BPXR lateral skull view to accurately characterize facial morphology.

METHOD

A total of 17 landmarks and 11 angles were placed and/or calculated on lateral BPXR and lateral cephalograms of 13 patients by three investigators. Five methods of angle identification were performed: the direct construction of straight lines on lateral cephalograms (LC-A) and on BPXR (BPXR-A), as well as the calculation of angles based on landmark identification on lateral cephalograms (LA-L) and on BPXR with the PA image (BPXR-LPA) or without (BPXR-L). Intra- and interoperator reliability of landmark identification and angle measurement of each method were calculated. To determine the most reliable method among the BPXR-based methods, their concordance with the reference method, LC-A, was evaluated.

RESULTS

Both imaging techniques had excellent intra- and interoperator reliability for landmark identification. On lateral BPXR, BPXR-A presented the best concordance with the reference method and a good intra- and interoperator reliability.

CONCLUSION

BPXR provides a lateral view of the skull suitable for cephalometric analysis with good reliability.

Relationship between European postgraduate programme accreditation and national research output: The case of the Network of Erasmus-Based European Orthodontic Postgraduate Programmes (NEBEOP) in orthodontics. A bibliometric study

AIMS

To assess in each European country the correlation between the number of Network of Erasmus-Based European Orthodontic Postgraduate Programmes (NEBEOP) members and orthodontic research activity. Secondary objectives were to describe and quantify Europe's orthodontic research.

MATERIALS AND METHODS

Articles published between 2014 and 2018 in 4 major orthodontic journals (American Journal of Orthodontics and Dentofacial Orthopedics, European Journal of Orthodontics, The Angle Orthodontist, Orthodontics and Craniofacial Research) and oral presentation abstracts of five European Orthodontic Society (EOS) congresses were analysed. For each European country, the total number of orthodontic programmes and NEBEOP memberships were collected. Descriptive statistics were performed, and Spearman correlation coefficients and risk ratios were calculated.

RESULTS

2039 articles and 261 oral presentation abstracts were included. Correlation coefficients between national number of publications, oral presentations, sum of these, all adjusted for population, and number of NEBEOP members in each country were 0.64, 0.65 and 0.62, respectively. Risk ratios were all above 1 and statistically significant for number of NEBEOP memberships per country, indicating positive associations with national orthodontic research productivity. Europe accounted for 30.5% of publications and 68.6% of oral presentations at EOS congresses during this period. European orthodontic research was not evenly distributed, since 9 countries were responsible for around 80% of the output.

CONCLUSIONS

A positive association was found between number of NEBEOP programmes and national research activity. These results could be an additional argument to support similar pan-European initiatives and guidelines for postgraduate education, not only in orthodontics but in all other dental specialties.

Voxel-based superimposition of Cone Beam CT scans for orthodontic and craniofacial follow-up: Overview and clinical implementation

INTRODUCTION

The increasing use of three-dimensional (3D) imaging in orthodontics has led to the development of 3D superimposition techniques. These techniques use stable anatomic structures as references in order to compare Cone Beam CT (CBCT) scans of the same subject at different time-points. Three methods have been described in the literature: landmark-based, surface-based and voxel-based 3D superimpositions.

OBJECTIVE

This article focuses on the voxel-based approach, which is the most described and the only one that can be fully automatized. The aim of this paper is to offer clinicians a practical tutorial on craniofacial voxel-based 3D superimposition.

MATERIAL AND METHODS

We provide an updated overview of the available implementation methods, describing their methodology, validations, main steps, advantages and drawbacks. The historical open-source method is the most widespread for research purposes, but takes around three hours to achieve for an experienced operator. Several commercially-available software perform superimpositions in a few minutes.

RESULTS

We used two of the available methods to conduct the superimposition process with three representative clinical cases in order to illustrate the different types of results that can be obtained.

CONCLUSIONS

Commercially-available software provide user-friendly and fully automatized superimposition methods, allowing clinicians to perform it easily and helping to reduce human error in image analysis. Still, quantitative evaluation of the results remains the main challenge of this technique.