First and second images obtained from first and second imaging modalities, respectively, are set as a target image and an object image, respectively. The object image is segmented into one or more anatomic segments. Each segment is associated with a respective anatomic class. At least one attribute is assigned to at least one of the anatomic segments based on the anatomic class corresponding to said at least one anatomic segment. A registration is performed with the object image and the target image, wherein the registration is constrained by the assigned attribute(s).
