Numerous diseases find their origin and their diagnosis in the physiological behavior of vascular networks [10]. In particular, to understand the architecture and growth of a tumor the study of blood flows is crucial. Recently, Ultrasensitive Doppler has enabled 4D ultrasound imaging of tumor micro-vasculature in mice [5]. In this study, we propose new computational tools to monitor the growth of a tumor vascular network by registering in time and space this new highly sensitive temporal data. We first quantify the acquired data using the minimal-path based framework we introduced in [4]; the vascular network paths around the tumor are segmented from images obtained for four days of observation; local geometrical parameters such as diameters are also estimated. Then, using a point cloud representation of the segmented vascular networks, we develop point cloud registration algorithms that automatically align similar vascular structures, thus allowing a better visualization of the growth and the evolution of the tumor vascular network. A rigid registration model is first considered by manually selecting similar features from two temporal different observations of the tumor. More accurate results are then obtained by automatically extracting invariant vascular patterns. Finally, combining rigid transformations to non-linear deformation models produce a very accurate time matching between invariant vascular structures.