@article{BrollRosentrittSchlegletal., author = {Broll, Alexander and Rosentritt, Martin and Schlegl, Thomas and Goldhacker, Markus}, title = {A data-driven approach for the partial reconstruction of individual human molar teeth using generative deep learning}, series = {Frontiers in Artificial Intelligence}, volume = {7}, journal = {Frontiers in Artificial Intelligence}, publisher = {Frontiers Media SA}, issn = {2624-8212}, doi = {10.3389/frai.2024.1339193}, abstract = {Background and objectiveDue to the high prevalence of dental caries, fixed dental restorations are regularly required to restore compromised teeth or replace missing teeth while retaining function and aesthetic appearance. The fabrication of dental restorations, however, remains challenging due to the complexity of the human masticatory system as well as the unique morphology of each individual dentition. Adaptation and reworking are frequently required during the insertion of fixed dental prostheses (FDPs), which increase cost and treatment time. This article proposes a data-driven approach for the partial reconstruction of occlusal surfaces based on a data set that comprises 92 3D mesh files of full dental crown restorations.MethodsA Generative Adversarial Network (GAN) is considered for the given task in view of its ability to represent extensive data sets in an unsupervised manner with a wide variety of applications. Having demonstrated good capabilities in terms of image quality and training stability, StyleGAN-2 has been chosen as the main network for generating the occlusal surfaces. A 2D projection method is proposed in order to generate 2D representations of the provided 3D tooth data set for integration with the StyleGAN architecture. The reconstruction capabilities of the trained network are demonstrated by means of 4 common inlay types using a Bayesian Image Reconstruction method. This involves pre-processing the data in order to extract the necessary information of the tooth preparations required for the used method as well as the modification of the initial reconstruction loss.ResultsThe reconstruction process yields satisfactory visual and quantitative results for all preparations with a root mean square error (RMSE) ranging from 0.02 mm to 0.18 mm. When compared against a clinical procedure for CAD inlay fabrication, the group of dentists preferred the GAN-based restorations for 3 of the total 4 inlay geometries.ConclusionsThis article shows the effectiveness of the StyleGAN architecture with a downstream optimization process for the reconstruction of 4 different inlay geometries. The independence of the reconstruction process and the initial training of the GAN enables the application of the method for arbitrary inlay geometries without time-consuming retraining of the GAN.}, language = {en} } @article{BrollGoldhackerHahneletal., author = {Broll, Alexander and Goldhacker, Markus and Hahnel, Sebastian and Rosentritt, Martin}, title = {Generative deep learning approaches for the design of dental restorations: A narrative review}, series = {Journal of Dentistry}, volume = {145}, journal = {Journal of Dentistry}, publisher = {Elsevier}, issn = {0300-5712}, doi = {10.1016/j.jdent.2024.104988}, pages = {10}, abstract = {Objectives: This study aims to explore and discuss recent advancements in tooth reconstruction utilizing deep learning (DL) techniques. A review on new DL methodologies in partial and full tooth reconstruction is conducted. Data/Sources: PubMed, Google Scholar, and IEEE Xplore databases were searched for articles from 2003 to 2023. Study selection: The review includes 9 articles published from 2018 to 2023. The selected articles showcase novel DL approaches for tooth reconstruction, while those concentrating solely on the application or review of DL methods are excluded. The review shows that data is acquired via intraoral scans or laboratory scans of dental plaster models. Common data representations are depth maps, point clouds, and voxelized point clouds. Reconstructions focus on single teeth, using data from adjacent teeth or the entire jaw. Some articles include antagonist teeth data and features like occlusal grooves and gap distance. Primary network architectures include Generative Adversarial Networks (GANs) and Transformers. Compared to conventional digital methods, DL-based tooth reconstruction reports error rates approximately two times lower. Conclusions: Generative DL models analyze dental datasets to reconstruct missing teeth by extracting insights into patterns and structures. Through specialized application, these models reconstruct morphologically and functionally sound dental structures, leveraging information from the existing teeth. The reported advancements facilitate the feasibility of DL-based dental crown reconstruction. Beyond GANs and Transformers with point clouds or voxels, recent studies indicate promising outcomes with diffusion-based architectures and innovative data representations like wavelets for 3D shape completion and inference problems. Clinical significance: Generative network architectures employed in the analysis and reconstruction of dental structures demonstrate notable proficiency. The enhanced accuracy and efficiency of DL-based frameworks hold the potential to enhance clinical outcomes and increase patient satisfaction. The reduced reconstruction times and diminished requirement for manual intervention may lead to cost savings and improved accessibility of dental services.}, language = {en} } @article{BrollGoldhackerHahneletal., author = {Broll, Alexander and Goldhacker, Markus and Hahnel, Sebastian and Rosentritt, Martin}, title = {Morphological effects of input data quantity in AI-powered dental crown design}, series = {Journal of dentistry}, volume = {159}, journal = {Journal of dentistry}, publisher = {Elsevier}, doi = {10.1016/j.jdent.2025.105767}, pages = {1 -- 8}, abstract = {OBJECTIVES: This retrospective in vitro study evaluated the impact of input data quantity on the morphology of dental crowns generated by AI-based software. The hypothesis suggests that increased input data quantity improves the quality of generated occlusal surfaces. METHODS: A dataset comprising n=30 patients (11 males, 19 females; age: 22-31 years) was analyzed. Input data was categorized into full dentition (full), quadrant data (quad), and adjacent teeth (adj). AI-based software (Dentbird Crown, Imageworks Inc.) generated crowns for a single lower first molar (36/46). Metrics were proposed to assess the morphology and occlusal relationships of the crowns, with the original tooth as reference. STATISTICS: Friedman Chi-Square tests, Wilcoxon signed rank tests, Kendall correlation and Fligner-Killeen tests (α = 0.05). RESULTS: Full and quad groups provided consistent reconstruction quality with no significant differences in morphology and occlusal relationships. The adj group showed significant (p<0.05) morphological deviations and higher reconstruction failure rates compared to the full and quad groups. Correlations (median: 0.19; min-max range: 0.01-0.54) indicate that the proposed metrics capture distinct morphological and functional crown aspects. CONCLUSION: The software reliably reconstructed crowns with at least quadrant-level input data. Performance declined with reduced input. Full-jaw scans did not enhance accuracy compared to quadrant data. CLINICAL SIGNIFICANCE: Increased input data quantity can improve the accuracy of AI-based restorations. As a result, prosthodontists benefit from predictable, accurate restoration proposals that reduce the need for digital chairside adjustments as well as manual modifications after fabrication. This streamlines clinical workflows and enhances the quality of restorations. Quadrant-level data has proven sufficient to generate high-quality reconstructions. Further input data did not significantly improve the accuracy of the reconstructions. The proposed metrics enable quantitative assessments of morphological and functional restoration quality, supporting reliable AI-driven workflows.}, language = {en} } @article{BrollHahnelGoldhackeretal., author = {Broll, Alexander and Hahnel, Sebastian and Goldhacker, Markus and Rossel, Jakob and Schmidt, Michael and Rosentritt, Martin}, title = {Influence of digital crown design software on morphology, occlusal characteristics, fracture force and marginal fit}, series = {Dental Materials}, journal = {Dental Materials}, publisher = {Elsevier}, issn = {0109-5641}, doi = {10.1016/j.dental.2025.09.003}, pages = {8}, abstract = {Objectives The study evaluated the influence of digital design software on crown morphology, occlusal characteristics, fracture force, and marginal fit across varying preparation designs for an identical target tooth. Methods A resin-based tooth (tooth 36) was digitized, manufactured ( ), individually prepared and re-digitized. Five design groups were established using conventional software proposals, technician designs, two AI-based software solutions, and natural tooth-based reference designs. All systems employed consistent parameters. Crown designs were digitally assessed using quantitative morphological and occlusal metrics in reference to the original tooth. Crowns were milled, marginal fit was measured via digital microscopy, and fracture resistance was determined after thermal cycling and mechanical loading. Results Morphological metrics revealed statistically significant deviations across groups, with the technician design achieving the best performance. Occlusal metrics showed high deviations in the positional accuracy of the contact points across all groups. Technician and AI-based designs exhibited comparable functional results. None of the design groups were able to achieve contact with all relevant antagonist teeth, due to high deviations in the mesiolingual cusp. Conventional software designs exhibited the lowest fracture forces. Significant improvements were achieved through technician intervention. Vertical marginal discrepancies remained comparable across groups. Significance Improved functional and morphological design combined with high fracture resistance can reduce the need for clinical adjustments, minimize wear, and enhance crown longevity. Digital design software significantly influences crown morphology, occlusal characteristics and fracture forces. Vertical marginal discrepancies remain similar. AI-driven approaches demonstrate comparability with technician designs in terms of fracture forces, functional performance, and marginal fit.}, language = {en} }