GPU accelerated FEM for simulation and segmentation

• 

  Project Week Main Page

• 

  FEM deformation

Key Investigators

• Megumi Nakao and Nobuhiko Hata

Objective

We are developing a fast FEM-based mesh deformation library for interactive simulation and segmentation. The goal in this project week is summarized as

• integrate linear FEM-based deformation algo into the slicer
• perform deformation on specific data
• implement GPU-based acceleration for real-time deformation

Approach, Plan

• integrate a liner FEM solver(original C++ source) into the 3D slicer framework.
• make tetrahedral meshes from medical images and set up physical conditions. Other mesh formats, points and surface models will be probablly accepted.
• evaluate some deformation results qualitatively and improve the simulation.
• try GPU-based acceleration by modifying the linear equation solver using CUBLAS library.

Progress

I will work in Wendnesday, Thursday afternoon and Friday.

References

• M. Nakao and K. Minato, "Physics-based Interactive Volume Manipulation for Sharing Surgical Process", IEEE Trans. on Information Technology in Biomedicine, 2009. (To appear)
• M. Nakao, T. Kuroda, M. Komori, H. Oyama, K. Minato and T. Takahashi, "Transferring Bioelasticity Knowledge through Haptic Interaction", IEEE Multimedia, Vol. 13, No. 3, pp.50-60, Jul. 2006.
• My homepage: http://kotaro.naist.jp/~meg/eindex.html