KEYNOTE 

We will present several physics-based modeling methods suitable for the modeling and analysis of volumetric objects. 

We will first present our work for the volumetric modeling and analysis of the heart from MRI-tagged data. By constructing a volumetric model of the Left- and Right Ventricles (LV-RV) from contour and volumetric tagged data we estimate the shape and motion of the heart during systole and diastole. In addition based on the use of the Navier Stokes equation we simulate the blood flow in the LV. We demonstrate quantitative analysis of heart motion for normal and pathological cases.

Then we will present novel volumetric segmentation methods based on the integration of 3D deformable models and Markov Random Fields. These hybrid segmentation techniques couple low level image processing methods with higher level model-based approaches and lend themselves naturally to minimal user interaction and more robust segmentation.

Finally we will present our efforts towards modeling the 3D motion of fluids based on the use of the Navier-Stokes equations. We will demonstrate modeling results of both fluids and gases.