Model-Based Analysis of Cardiac Motion from Tagged MRI Data

Applications of Computer Vision in Medical Image Processing

Volume

Papers from the 1994 AAAI Spring Symposium

Contents

In this paper we present a new technique that describes the time-varying shape, deformation and shape of the LV in terms of a few "global" parameter functions, such as twist, whose value is allowed to vary locally. In this way the complex motion of the heart may be described by the same small number of parameters, which vary from region to region. Furthermore, these parameters are intuitive and can be used by a physician without further complex processing. Our approach is based on the development of a new family of parameterized deformable primitives suitable for this application. These deformable primitives are parameterized using global parameter filnctions whose value varies across the primitive’s shape as opposed to being constant [14, 9]. Through the use of appropriate parameterization the axes of our new deformable primitives can be non-straight. This is a lnajor generalization compared to parameterized primitives such as superquadrics, cylinders and cubes, comlnonly used in the vision literature. Furthermore, generalized cylinders [4, 7] even though they allow shapes with nonstraight axes, do not offer shape representation in terms of a few parameters.