3D reconstruction from medical images

To build patient-specific models of the anatomy, I am interested in image registration techniques based on elastic deformable models. The approach is to iteratively align a generic 3d model with the structures of interest in the images. To improve robustness, models need to be constrained with a priori information about shapes and allowed kinematics. I am exploring algorithms to apply such constraints, mainly in the context of human musculoskeletal system 3d reconstruction, and MRI imaging.

 

Simulation of deformable solids

 

This research deals with the development of new algorithms for the mechanical simulation of deformable solids in real-time. Most of the existing approaches based on continuum mechanics, either mesh-based or mesh-less, use dense discretizations to deal with large non-linear deformations and complex material distributions. To reduce computations for real-time applications, I am studying coarser (adaptive) discretizations and techniques to embed pre-computed or analytic solutions into the kinematic model.