Model-to-volume registration for endovascular aneurysm repair

Endovascular aneurysm repair involves catheter navigation through the vasculature to the treatment site under live imaging guidance, which can be greatly facilitated by smart visualization technology involving the pre-operative 3D CT volume. However, the pre-operative data first needs to be aligned with the patient by registering it with an intraoperative non-contrast-enhanced Cone-beam CT (CBCT), which typically has a relatively low signal-to-noise ratio. In this paper, a fully automatic rigid-body model-to-volume registration algorithm is proposed to register the 3D mesh model extracted from CT with the noisy volumetric CBCT data. Novel similarity measures between the model and the volume are proposed for highly efficient registration involving global optimization. The proposed method was validated on 9 real datasets from EVAR patients. Results show that the proposed algorithm is highly accurate, robust and computationally efficient.