In this paper we propose a registration method that combinesintensity information with geometrical information inthe form of curves and surfaces derived from lung CT images.Vessel tree centerlines and lung surfaces were extractedfrom segmented structures. First, a current-based registrationwas applied to align the pulmonary vessel tree and the lungsurfaces. Subsequently, the resulting deformation field wasused to constrain an intensity-based registration method. Weapplied the combined registration on a set of image pairs, extractedat the end exhale and the end inhale phases of 4DCTscans. The proposed combined registration was comparedto intensity-based registration, using a set of manuallyselected landmarks. The proposed registration decreases themean and the standard deviation of the target registration errorsfor all 5 cases to on average 1:47 +/- 1:05 mm, compared tothe intensity-based registration without constraint 1:74 +/- 1:31mm.