Abstract:
3D-reconstruction of organ deformations from ultrasound image
sequences is an active goal for online medical imaging.
We focus here on "speckle tracking" methods in intracardiac
echocardiography : after patient's catheterization, sequences of
echographic 2D-images are acquired to successively display views of
"parallel slices" of the cardiac muscle.
Algorithmic recovery of 3D non linear myocardial deformations must
overcome the speckle effect, which generates strong dynamic noise
textures, modeled by non Gaussian multiplicative noise, and which
decorrelate in time under large deformations.
Maximum likelihood formulations for the quality of elastic
registration of such image sequences enable algorithmic reconstruction
of 3D-cardiac motion, but may still be refined by more accurate
texture matching. Robust solutions for this ill-posed problem
combine probabilistic texture models with various regularization
strategies to recover diffeomorphic registration of image sequences.
Joint work with Yipeng Li (UH) and Dirar Koury (Methodist Hospital)
This seminar is easily accessible to persons with disabilities. For more information or for assistance, please contact the Mathematics Department at 743-3500.