EUVIMEDSkopri EUVIMED
Crossref

Machine‐Learning‐Based Spectral Modeling: A Biomimetic Guide for Enhancing Esthetics

L J Herrera iD, R Ghinea iD, M M Perez iD, R D Paravina iD

Journal of Esthetic and Restorative Dentistry · 2025

DOI: 10.1111/jerd.13380 ↗

At a glance

Journal article in Journal of Esthetic and Restorative Dentistry, 2025. Written by L J Herrera and 3 co-authors. Institution: Department of Computer Engineering, Automatics and Robotics, School of Technology and Telecommunication Engineering, University of Granada, Granada, …. Major MEDLINE topics: Machine Learning, Esthetics, Dental, Biomimetics, Prosthesis Coloring. Cited 4 times so far (Crossref); 25 references. Funded by: Spanish National Plan for Scientific and Technical Research and Innovation, Ivoclar Vivadent AG, Ivoclar Vivadent.

AstrattSubjectsDettalji biblijografiċiCiteRelated studies

Abstract (original language)

Objective

To provide guidelines and means for optimal coverage and distribution of computer models with 1–10 clusters, designed based on an in vivo extensive dental colorimetric database and compare the findings with some reputable shade guides. Materials and Methods The Fuzzy C‐Means (FCM) algorithm was used to optimize the tab position in the CIELAB color space, while MATLAB Fuzzy Logic was used to implement the codes. A minimization process was performed to fine‐tune the cluster centers, minimizing Coverage Error ( CE 00 ) and Maximum Error ( ME 00 ). Spectral reconstruction was performed, and the results were compared with the corresponding data for VITA classical A1–D4 (VC), Linearguide 3D‐Master (LG), and Bleachedguide 3D‐Master (BG) shade guide. Paired t ‐test was employed to assess the significance of CE 00 differences.

Results

CE 00 and ME 00 ranges for 1‐to‐10‐tab models were 3.8 (14.2) to 1.7 (4.3), while the corresponding values for shade guides were 3.1 (9.1) for VC, 2.3 (6.0) for LG, and 2.9 (9.3) for BG. Hence, the 4‐tab‐model with CE 00 of 2.2 and ME 00 of 7.1, outperformed CE 00 of LG (the lowest/best CE 00 among shade guides), while the 6‐tab‐model, with CE 00 of 2.0 and ME 00 of 5.6, outperformed its ME 00 . A paired t ‐test confirmed that increasing the number of tabs models resulted in significant CE improvement in all cases ( p ≤ 0.05).

Conclusions

Shade guide models with only 4 to 6 tabs, optimized using computational techniques, outperformed physical dental shade guides tested, exhibiting significantly lower (better) Coverage Error and Maximum Error. Clinical Significance Given that the spectral modeling was performed on an extensive in vivo database of natural teeth, this approach was validated as a biomimetic guide for shade guides and corresponding materials of the future, offering a possibility for better shade matching with fewer shades/shade tabs.

Subjects

MeSH terms (MEDLINE)

Machine LearningEsthetics, DentalBiomimeticsProsthesis ColoringDental Prosthesis DesignHumansColorimetryAlgorithmsColorFuzzy Logic

Author keywords

Fuzzy C‐Means (FCM) algorithmcolordental shade guidesmachine learningspectral modeling

Authors and institutions

  • Department of Computer Engineering, Automatics and Robotics, School of Technology and Telecommunication Engineering, University of Granada, Granada, Spain.
  • Faculty of Sciences, Department of Optics, University of Granada, Granada, Spain.
  • John M. Powers Ph.D., Center for Biomaterials and Biomimetics (PCBB), UTHealth Houston School of Dentistry, Houston, USA.
  • Department of Restorative Dentistry and Prosthodontics, UTHealth Houston School of Dentistry, Houston, USA.

ORCID profiles: L J Herrera, R Ghinea, M M Perez, R D Paravina

Funding

  • Spanish National Plan for Scientific and Technical Research and Innovation
  • Ivoclar Vivadent AG
  • Ivoclar Vivadent

Cite

APA

Herrera, L. J., Ghinea, R., Perez, M. M., & Paravina, R. D. (2025). Machine‐Learning‐Based Spectral Modeling: A Biomimetic Guide for Enhancing Esthetics. Journal of Esthetic and Restorative Dentistry. https://doi.org/10.1111/jerd.13380

Vancouver

Herrera LJ, Ghinea R, Perez MM, Paravina RD. Machine‐Learning‐Based Spectral Modeling: A Biomimetic Guide for Enhancing Esthetics. Journal of Esthetic and Restorative Dentistry. 2025. doi:10.1111/jerd.13380

BibTeX
@article{Herrera2025,
  author = {Herrera, L J and Ghinea, R and Perez, M M and Paravina, R D},
  title = {Machine‐Learning‐Based Spectral Modeling: A Biomimetic Guide for Enhancing Esthetics},
  journal = {Journal of Esthetic and Restorative Dentistry},
  year = {2025},
  doi = {10.1111/jerd.13380}
}
Start for free →