Date: October 7, 2011
Time: 10:00 AM
Place: E & CS Auditorium, First Floor 


Title:

Real-Time Image-to-Mesh Conversion

Abstract:

The use of medical images provides rapidly growing amounts of data facilitating 
personalized diagnostics and treatment. For example, non-rigid registration of 
pre- and intra-operative images is critical in the operation of guided 
navigation systems. Another example is the development of interactive surgery 
simulators which, in conjunction with a haptic and a visual computer interface, 
help to train young surgeons and has been shown to have a measurable impact on 
surgical skill and patient outcome. These and other similar applications 
involve modeling of physical phenomena by solving systems of partial 
differential equations (PDEs). When PDEs are defined over geometrically complex 
domains, they often do not admit closed form solutions. In these cases, the 
PDEs are solved approximately using discretizations of domains into simple 
elements like triangles in two dimensions, and tetrahedra in three dimensions. 
These discretizations are called finite element meshes.

In this talk I will describe key requirements, recent progress, and open 
opportunities in real-time image-to-mesh conversion. Many applications, for 
example real-time computer assisted surgery, or medical surgery simulators 
impose time and/or mesh size constraints that cannot be met on a single 
sequential machine. As a result, the development of parallel mesh generation 
algorithms is required. In addition, the requirements on shape, size, and 
location of mesh elements impose additional and often conflicting constraints. 
Being able to guarantee these properties of the resulting meshes is critical 
for the robustness and usability of the software, and the problem is 
exacerbated when parallel machines are used. I will describe the recent 
progress in addressing these requirements using a novel parallel generalized 
Delaunay refinement approach exploited on multiple levels of concurrency, as 
well as a novel lattice decimation approach. I will also briefly outline the 
current and emerging multi-area collaborations that strengthen this research 
and expand its impact.


Bio:

Andrey Chernikov graduated with a Ph.D. in Computer Science from the College of 
William and Mary in 2007 under the guidance of Prof. Nikos Chrisochoides. His 
dissertation work "Parallel Generalized Delaunay Mesh Refinement" was awarded a 
Distinguished Dissertation Award in the Natural and Computational Sciences by 
the Graduate Studies Advisory Board, College of William and Mary. Prior to his 
Ph.D. work he obtained M.S. and B.S. degrees with distinction in Applied 
Mathematics and Computer Science from the Kabardino-Balkar State University in 
Russia. He was also twice named a Soros Student. After the Ph.D. work he had 
been a Visiting Assistant Professor and a Postdoctoral Associate in Computer 
Science at William and Mary. From Fall 2010 he has been working as a Research 
Assistant Professor in Computer Science at ODU. His research interests include 
image analysis in medical and bio-material modeling and simulation, quality 
mesh generation, high-performance scientific computing, and their applications.