Genetic algorithm fitting of deformable superquadrics applied to left ventricle visualization

For many years, reconstructing 3D objects has been of great interest in scientific imaging. The most common techniques reconstruct rigid objects using simple forms that do nor permit the natural movement of objects. This paper presents the study of a surface very useful in modelling natural 3D objects, deformable superquadrics. The principal feature of this kind of surface is that based on a small set of parameters, various anatomical shapes and their intrinsic movements can be modelled. Superquadric parameters are estimated using genetic algorithms to minimise the distance of the model from the set of data that comes through ultrasound from the organ being modelled