Scholarlink

Computationally modelling the soft-tissue changes caused by orthodontics is challenging due to the small displacements caused by the treatment. Large precision is required for geometric modelling, and consequently during data collection. However, data gathered before and after treatment contains spurious facial changes from many sources, such as aging or weight-change. In this work, a novel data-collection method based on specialised dental obturators is used to obtain a clean dataset without confounding factors. The data quality is demonstrated by comparison to a cone-beam CT (CBCT) dataset of treatment outcomes. Using this dataset, we evaluate an FEM simulation with models of contact, jaw motion, and active-lip deformation, performing ablation studies to assess their impact on accuracy. We also show how differentiable simulation can optimize material parameters in the implemented models. Our results demonstrate the utility of in-vivo simulated data, and the evaluation of the simulator highlights the role of the jaw and lip in contact-driven facial deformation.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating a Differentiable Facial FEM Model with In-Vivo Simulated Data for Orthodontic Treatment Outcome Prediction

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Daniel Dorda,
  • Daniel Peter,
  • Andrey Latyshev,
  • Luís Azevedo,
  • Niko Ben Huber,
  • Baran Gözcü,
  • Irena Sailer,
  • Markus Gross,
  • Bernhard Thomaszewski,
  • Barbara Solenthaler

摘要

Computationally modelling the soft-tissue changes caused by orthodontics is challenging due to the small displacements caused by the treatment. Large precision is required for geometric modelling, and consequently during data collection. However, data gathered before and after treatment contains spurious facial changes from many sources, such as aging or weight-change. In this work, a novel data-collection method based on specialised dental obturators is used to obtain a clean dataset without confounding factors. The data quality is demonstrated by comparison to a cone-beam CT (CBCT) dataset of treatment outcomes. Using this dataset, we evaluate an FEM simulation with models of contact, jaw motion, and active-lip deformation, performing ablation studies to assess their impact on accuracy. We also show how differentiable simulation can optimize material parameters in the implemented models. Our results demonstrate the utility of in-vivo simulated data, and the evaluation of the simulator highlights the role of the jaw and lip in contact-driven facial deformation.