Introduction: Typically, image fusion (IF) of preoperative 3D image data sets with x-ray (XR) fluoroscopy involves one of the following techniques: a) image-based 3D-3D registration by means of cone-beam computed tomography (CBCT), or b) direct image-based 2D-3D registration, where two angiographic images acquired at least 30° apart are used for co-registration. Taking into account additional radiation exposure caused by CBCT and/or injection of additional volume of contrast agent (CA) required for visualization of the anatomy, further refinement of the IF procedure is required.
Objectives: The aim of this study was to demonstrate an approach to patient-specific co-registration of pre-interventional CT data set with real-time XR fluoroscopy during transfemoral transcatheter aortic valve implantation (TAVI), not requiring any additional CA or XR scan as compared to the conventional non-IF procedure.
Methods: 3D models of the abdominal aorta with main trunks of the left and right coronary arteries and main artery branches, left and right iliofemoral arteries, hip bones and left ventricle were generated, yielding a highly detailed anatomical model of the target structures (Figure A). To link the model to the x-ray system geometry co-registration was initially performed in the iliofemoral region using routinely acquired arteriographies. The hip bones served as additional markers during alignment of the overlay. On-time refinement of the co-registration was then performed during the on-going procedure yielding accurate superposition of the anatomic model at each time point. In total 53 procedures were performed with IF and its outcome was compared with a control group made up of 53 randomly selected procedures without IF matched by the number of TAVI prosthesis type used in the IF group. Performance of the procedures with IF and without IF was evaluated quantitatively in terms of procedural characteristics and technical success of device placement.
Results: The described co-registration approach allowed to overlay the patient-specific anatomic model onto the live XR images and provided individual 3D information on the substrate during the entire TAVI procedure. This allowed to safely guide the intervention including puncture of the common femoral artery above the femoral bifurcation on the sheath side (Figure B) and sheath introduction, placement of the double-filter cerebral embolic protection device (Figure C) and deployment of the valve prosthesis (Figure D). Significant reduction of the CA volume [90 (50–90) vs. 100 (80–100) ml, p < 0.001], overall procedure time [55 (42–68) vs. 63 (59–72) min, p < 0.001] and fluoroscopy time [926 (765–1309) vs. 1121 (967–1271) s, p = 0.0218] was achieved with IF, and the radiation dose could be lowered [53 (38–79) vs. 58 (44–79) Gy cm2, p = 0.3045]. Successful device placement was achieved in 51 (96%) cases in the IF group [vs. 49 (92%) in control group, p = 0.8859].

Conclusions: A co-registration approach for IF making use of routinely acquired fluoroscopic XR images and not requiring any additional CA or CBCT scan for co-registration is introduced.  IF supports the TAVI procedure from its initial stage (starting with the puncture of the femoral artery) by providing 3D anatomy of the target structures over the entire course of the intervention. Providing significant reduction of contrast volume, fluoroscopy and procedure time, it indicates substantially improved performance of the procedure.