Multimodal Medical Volume Registration Based on Spherical Markers

Abstract

We propose volume registration procedures based on spherical artificial markers presented in medical multimodal data sets (MRI and CT, especially). The procedures proposed are either semi-automatic or fully-automatic. A semi-automatic approach requires to label approximate locations of the spherical markers in the data sets and then registration operates autonomically. A fully-automatic approach does not require any user interaction, i.e. all registration subtasks – namely segmentation of spheres, finding the correspondence between two sets of spheres and, finally, computing geometrical transformation that maps the first set of spheres onto the second one – are performed automatically by the computer.