吕思哲（Keith）
Real Name: Lv, Sizhe
Nick Name: Keith.Lv
Hobbies: Comic, Movie and Running
Birthday: Feb 21st, 1985
E-mail: lvsizhe@gmail.com
Sizhe is a M.S. student of Shanghai Jiao Tong University (SJTU), majoring
Computer software and Applications, supervised by prof. Lixu Gu. She received
her B.S. degree in Software Engineering from Department of Software, Shanghai
Jiaotong University (SJTU) in 2007. Sizhe is now a core member of Image Guided
Surgery Therapy Lab (IGST) of SJTU. Her research interest includes Computational
Geometry, mesh resconstruction from medical images, dynamic mesh in virtual
surgery simulation, and computer graphics.
目录
|
Reserach && Projects
Mesh reconstruction from medical images
 |
 |
 |
 |
Reconstructing meshes from 3D medical images is an important but complicated task, which provides a fundamental data structure for Bio-medical applications. Instead of following the popular methods, we sample on the medical images with centroidal Voronoi tessellations, and then build the Delaunay mesh from the sampled points directly. By applying the algorithm on the CT images and evaluating the mesh quality metrics, the solution presents to be good at reconstructing complicated objects and generates optimized meshes, which shows that it is suitable for anatomical structure modeling.
Dynamic mesh
 |
 |
 |
 |
In the virtual surgery system, the quality of the meshes usually become worse and worse during the simulation, and the mesh quality is highly related to the accuracy, stability and efficiency of the virtual surgery systems. So, how to maintain the quality of the surface mesh is a critical and challenge work for a virtual surgery system. We develop a schema bases on loose r-sample theory which can effectively protect the surface meshes’ quality during the objects deformation. It's a very useful tool which allowing users to use point operations instead of mesh operations, delivering an uniform data structure to support the typical mesh operations and greatly reduces the difficulty of interactive algorithm design for virtual surgery systems.
Following is the list of the developed applications based on loose r-sample dynamic mesh schema.
Remeshing from existing surface mesh
 |
 |
 |
 |
The loose r-Sampled dynamic mesh schema can remesh from an existing object and improved the surface meshes's quality.
Left is the input surface mesh and right is the output remeshed surface.
Collision detection
If all the objects are management in a loose r-sample based scene manager, it's very easy to detect the possible collision vertex pairs from the Delaunay triangulation maintained by the scene manager. In the right image, the possible collision vertex pairs are tagged by the pink segments.
Mesh quality protection during object deformation
 |
 |
 |
 |
 |
 |
The quality of the static mesh will be worse and worse during the deformation (the second row), the loose r-sampled based dynamic surface will resample the surface mesh during the deformation and keep the quality of the meshes (the first row). In the above image, the red are bad triangles and the blue are good ones.
Drill an object
 |
 |
 |
 |
The loose r-sampled based schema can simulation the surgery operations. The above image shows the simulation of the drilling an object. Note that the surface is always kept in a high quality status.
Soft-tissue modeling
 |
 |
 |
Take part in the research of soft-tissue modeling (mainly FEM and Mass-Spring model) and develop the spine-modeling && craniofacial surgery project.
Publications
[1] Sizhe Lv, Xubo Yang, Lixu Gu, Xiaojin Xing, Lei Pan and Min Fang, “DelaunayMesh Reconstruction From 3DMedical Images Based On Centroidal Voronoi Tessellations”, The International Conference on Computational Intelligence and Software Engineering (CiSE), 2009. (pdf)
[2] Sizhe Lv, "The reconstruction and quality maintenance of dynamic meshes in virtual surgery simulator （虚拟手术中动态网格的重建及质量保持)", Master degree thesis. (pdf)