Jens Georg Schmidt, Guntram Berti, Jochen Fingberg, Junwei Cao, and Gert Wollny. A Finite-Element based Tool Chain for the Planning and Simulation of Maxillo-Facial Surgery. In P. Neittaanmäki and T. Rossi and K. Majava and O. Pironneau, editors, European Congress on Computational Methods in Applied Sciences and Engineering ECCOMAS 2004.
We describe a chain of computational tools for the simulation and planning of maxillo-facial surgery. Malformations of the midface such as maxillary retrognathia can be treated by distraction osteogenesis: A part of the maxilla is separated from the rest of the skull (osteotomy) and pulled to the appropriate position. For large displacements, the pulling procedure is implemented by a distraction device fixed to the skull and can last several weeks. Often, additional plastic surgery is necessary to achieve an esthetically pleasing result.
A critical point in this procedure is the prediction of the surgery outcome. In current clinical practice, planning is based on CT scans and the surgeon's experience. Our tool chain simulates the displacements of bones and soft tissue, and therefore provides the possibility to try and compare different surgical treatments in silico.
The first step of our simulation is the segmentation of a CT image of the patient's head into different anatomical structures. Currently, we work on supplementing the simple thresholding used to distinguish bone and soft tissue by a more sophisticated registration-based segmentation, which will allow to differentiate also fat and muscles.
A surface mesh of the bone is handed over to a cutting tool, which allows the surgeon to define arbitrary 3D cutting paths in an intuitive way, and to specify repositioning of bone components. This tool is crucial for the whole process; it contains visual feedback in order to verify separation of components.
Based on the geometric cuts, a volume mesh is produced from the image data and boundary conditions corresponding to the repositioning are included. This mesh is used to simulate the distraction process by our parallel finite element code, which is combined with a fast multigrid solver and offers a variety of material laws (linear elastic, hyperelastic, viscoelastic) as well as different finite element formulations (linear, quadratic, mixed elements of tetrahedral, hexahedral and pyramidal shape). Thus, the degree of accuracy can be varied, ranging from linear, static and very fast simulations, providing a quick but rough impression to the surgeon, up to very detailed ones that take into account non-linearities and time dependencies of the underlying problem and need several hours of computing time. The ability to switch between fast and accurate simulations satisfies both the need for fast turn-around times for planning, and reliable prediction of surgery results.
Our work is carried in the context of a larger project (GEMSS) which aims at making advanced simulation accessible to the medical practitioner. Here, accessibility is understood both in terms of lowering the technical barriers for the "simulation layman" and of providing transparent access to large remote computational resources. Therefore, much effort beyond a "minimalistic working example" is devoted to developing robust, ergonomic and reliable services, and to offering them to the medical community via a secure Grid infrastructure.
On the GEMSS website, there is a demonstration of the process.