Auditory-Visual Speech Processing 2005
British Columbia, Canada
There is an increased need to study the human oropharyngeal anatomy for research in speech production, feeding motion, and breathing activities. Computational anatomic models are also useful to conduct virtual experiments without extensive use of human subjects, to plan and train surgeries. Recent improved methods for fast and high resolution imaging of internal organs, such as, Magnetic Resonance Imaging (MRI), three-dimensional Ultrasound, and Computer Tomography (CT) combined with automatic image extraction techniques, dynamic modeling techniques, and increased computation capacity, enhance possibilities to create realistic computational models.
One central organ of the vocal tract is the tongue, which has been modeled using statistical parametric models [1, 2], explicit shape descriptions [3, 4], physiological models , and dynamical models (both spring-mass , and finite element models [7, 8, 9]). A recent survey  describes existing methods in detail. For many surgical simulation applications, tongue models need to support the following:
For these requirements, finite element modeling provides a good solution and has a long tradition in Engineering [11, 12]. So far, only non real-time solutions have been developed for tongue models [7, 8, 13, 9]. Recent developments in the fields of physical-based animation [14, 15] and surgical simulations  provide finite element algorithms which can run in real-time to provide plausible results even for large deformations. Compared to spring-mass models, they have better stability and accuracy.
In this work, we will create a real-time two and three dimensional finite element model of tongue tissues and demonstrate a complete work flow from image data extraction to model configuration to model validation in the framework of ArtiSynth .
Bibliographic reference. Vogt, Florian (2005): "Finite element modeling of the tongue", In AVSP-2005, 143-144.