Assistant Professor
Department of Computer Science
University of Pittsburgh
Pittsburgh, PA 15260-9161
Adjunct Assistant Professor
Robotics Institute, School of CS
Carnegie Mellon University
Pittsburgh, PA
|
Home Publications Teaching Service Contact |
Vita
My research centers around applications of computational modeling, visualization, computer graphics, and computer science to other scientific disciplines. I am looking for motivated students to work with me.
I received my Ph.D. in computer science at Brown University in May 2007. While at Brown, I was a member of the Graphics Group, and I worked under the supervision of David Laidlaw (my advisor), Joseph (Trey) Crisco, and John Hughes. My doctoral research explored novel computational modeling, visualization and analysis tools that are needed to model anatomical joints and their variation with disease progression.
I received a B.S. and an M.S. from the Computer Science Department of the Politehnica University of Bucharest, in Romania. I spent there an extra year as a lab instructor, and a summer as an intern at Philips Research in the Netherlands. Frans Peters of Philips bears responsibility for me switching from AI to graphics (hi, Frans, and hi, Cristian Giumale, you were my first mentors :-).
Research and projects
Research focus: computational modeling, scientific visualization, computer graphics, computer vision, bioengineering, medical imaging and image processing, dynamic data-driven application systems.
|
Dynamic data-driven biological systems (Predictive computational models)
My PhD work introduced a framework for integrating measured data - such as medical images, tracked motion, and anatomy-book knowledge - into the physically-based simulation of anatomical joints. The resulting models and simulations can be used to analyze medical data collected from a specific individual. My current work in this direction pursues dynamic, data-driven models of biological systems. In the dynamic data-driven paradigm, the simulations and measurements become a symbiotic feedback control system. Dynamic models have enormous predictive power. Such models can be used to plan surgical interventions and to generate realistic computer animations. The development of predictive biological models opens further research directions I'm interested in. For example, what level of anatomical realism is necessary or useful when we design orthopedic implants and robotic arms, versus animating Pixar-like characters? |
|
Visual analysis tools (Discovering the unexpected) The goal of computational modeling is not only to detect expected events, such as might be predicted by models, but also to help users discover the unexpected - the surprising anomalies, patterns, or relationships that are then examined and assessed to develop new insight. Visualization is a strong tool in this sense, in particular when tightly coupled with modeling, simulation and interactive thinking techniques. In TVCG'07 and JOR'07 I presented JointViewer, a system for the visual exploration and analysis of joint biomechanics. JointViewer integrates modeling with visualization and simulation. It thus provides computational steering and helps guide investigations. The system is routinely used at Brown in both the visualization research and the bioengineering lab. I continue to pursue tools for the visual mining of complex and dynamic data. |
 
|
Image-based measurements (Image-based
motion tracking and estimating the
unmeasurable) Imaging allows us to measure not only biological shapes, but also their motion and deformation. Such information is essential in computational modeling. In TMI'06 I used unsupervised tissue classification to reduce imaging artifacts. As a result, I was able to develop a subvoxel-accurate method for tracking bone-motion from sequences of volume images. Results showed expanded capabilities and average accuracy improvements of 74% over the previous-best technique. With appropriate tools, models of small structures that are currently not measurable directly can be inferred from medical image data. In TBME'03 and EMBC'06 I introduced non-invasive methods for estimating individual-specific soft-tissue geometry solely from kinematic data and bone surfaces. The approach enhances the capabilities of radiation-based imaging: we are able to capture in vivo soft-tissue information that used to be measurable only through invasive means. |