Title: Hybrid Segmentations Methods, Organ Reconstruction and Tissue Modeling


                            Dimitris Metaxas
         Center for Computational Biomedicine, Imaging and Modeling
            Depts of Computer Science and Biomedical Engineering
                           Rutgers University
			email: dnm@cs.rutgers.edu
     
Abstract:
 We have recently been working on the next generation of segmentation 
methods which combine statistical
methods and Deformable models. The first class of these methods is based 
on the use of graphical models to integrate
MRFs and Deformable models, while the second is a novel class of 
statistical models that integrate shape and texture.
 We present the use of these methods for the automated segmentation of 
noisy and fuzzy organ boundaries such as those of the
heart and  the detection of cancer. Finally, we present our methodology 
for the heart's  motion and shape analysis which includes the
incorporation of tissue properties and fiber orientation.