05-0018 Deformable vessel-based registration

Blood vessels in the human body are organized as tree structures. Our technique for deformable vessel-based registration relies on this tree configuration when registering medical images. The method for this registration system involves aligning a source image with the target image by registering a model of the tubes in the source image directly with the target image. Our innovative approach is different from traditional approaches by (1) combining both rigid and deformable transformations in a hierarchical manner, (2) combining geometry and intensity information, and (3) operating as an instance of model-to-image registration. This technique offers accuracy and efficiency, as it is 10-20 times faster than traditional methods.


  • This model-to-image registrations shows sub-voxel accuracy
  • Robustness to noise and a convergence time of less than 1 minute

Contact Information

Name: Research Tool Manager

Email: researchtools@unc.edu

Phone: 919-966-3929