A method of object mapping in a multi-dimensional dataset, said method comprising using a processor to perform the steps of: segmenting the multi-dimensional dataset using multi-dimensional graphic objects, defining a set of geometric relations between a set of geometric templates using a geometrical relational application framework macro, associating the multi-dimensional graphic objects with the geometric templates, and constructing, for the multi-dimensional graphic objects, constituent models, wherein a constituent model is designated for segmenting a respective predetermined constituent structure, constructing a composite model for segmenting the multi-dimensional graphic dataset by determining at least two constituent structures that are incorporated in or related to the multi-dimensional graphic object, and forming the composite model based on respective constituent models that correspond to the respective determined constituent structures, the composite model being operative to segment the multi-dimensional graphic dataset by controlling the constituent models.