Estimation of 2D/3D rigid transformation for Computer-assisted Endovascular Navigation

Minimally invasive vascular interventions require the exact location and orientation of the endovascular tools (i.e., balloon, stent graft) with respect to the patient in the surgical environment. While a 3D image is often available preoperatively (typically an angio-CT scan) the practitioner mainly relies on the 2D images provided by the mobile fluoroscopic device (C-arm) to obtain a real-time feedback of the intravascular tool location as well as of the therapeutic actions. The main objective of this work is the study of an endovascular navigation system that enables the surgeon to improve the accuracy of the procedure along with a minimum use of the intra-operative 2D acquisitions, leading to less radiation exposure and less contrast agent injection. With the aim of exploiting the whole available information at the different observation stages through a cooperative approach, this paper focuses on the definition of the different rigid geometrical transformations implied in the 3D preoperative / 2D intraoperative images matching, as well as on the estimation of the transformation parameters based on the use and combination of different types of data and methods (calibration, 3D tracking, image registration).