A dynamic point matching for registration in image-guided minimally invasive vascular surgery: A simulation study

Registration is a key issue in image-guided minimally invasive vascular surgery (MIVS) when the electromagnetic device is used to tracking the catheter. We assume that the catheter advances in the vessel along the vascular centerline during MIVS. Based on this assumption, we proposed a dynamic point matching approach for registration, of which the transformation is estimated by aligning the point set of catheter's path to that of the vascular centerline. Due to the point set of the path obtained in real time, the point matching is performed dynamically. A matching function is introduced to indicate the correspondence between two point sets. The function is refined to find the optimal correspondence using Relax labeling, which account for local neighborhood structures. The transformation is estimated by minimizing the distance of the optimal corresponding points. A simulation experiment is designed and performed. The results show that the mean +/- standard deviation of target registration error is 2.72 +/- 0.03mm.