The invention relates to a method of segmenting a surface in a multi dimensional dataset comprising a plurality of images. In accordance with the method of the invention, at step 4 shape parameters and topology parameters for the object under consideration are acquired. Preferably that the multi-dimensional data set imaging the object is acquired at an acquisition step 1 and is subsequently stored in a computer-readable file 2. At step 5 the default shape parameters and topology parameters of a suitable segmentation algorithm 3 based on a deformable model are adapted with the value of the actual shape parameters and the topology parameters 4 for the given object. Subsequently, at step 6 the images constituting the multi-dimensional dataset are segmented using deformable model algorithm 6a with the adapted shape parameters and the adapted topology parameters yielding respective portions of the sought surface. After all sub-portions of the segmented surface are obtained for all images, the surface is tracked using per se known tracking algorithms 8a resulting in establishing spatial correspondence between said surface portions. Preferably, the method according to the invention is followed by the step of reconstructing 9 wherein for a given viewing angle the surface is reconstructed in virtual space. At step 11 the reconstructed surface is visualized on a suitable display means for user's analysis.