Image Fusion During Endovascular Aneurysm Repair, how to Fuse? An Overview of Registration and Implementation Strategies Plus Tips and Tricks

Introduction: The use of image fusion in the hybrid operating room is increasingly used. Image fusion enables physicians to deploy endovascular devices with a 3D roadmap of the vascular anatomy based on preoperative CTA or MRA. Previous studies described a decrease in nephrotoxic contrast volume, fluoroscopy time, radiation dose and procedure time in complex endovascular aortic repair (EVAR).1-4 However, there is no general consensus or guideline on the optimal technique and timing of image fusion. There are several options; 2D-3D bony landmark registration, 3D-3D aortic calcification registration, contrast enhanced cone beam CT (ceCBCT) registration. Moreover, registration can be done before or after the insertion of sheaths and (stiff) guidewires. The goal of this study was to determine image fusion accuracy. Methods: Several fusion strategies were analyzed. Strategy 1: 2D-3D registration before insertion of sheaths and guidewires. Strategy 2: 3D-3D registration before insertion of sheaths and guidewires. Strategy 3: 2D-3D registration after insertion of sheaths and guidewires. Strategy 4: 3D-3D registration after insertion of sheaths and guidewires. An overview is displayed in Figure 1A-1D. Strategy 1 was evaluated with clinical patient data. Strategies 2, 3 and 4 were evaluated with an infrarenal AAA phantom model with pelvis, vertebral column and renal calcifications as displayed in Figure 1E. For strategy 1, in total 11 EVAR patients (median age 75.5, all male) were included of which 4 were complex EVAR (fenestrated) and 7 standard EVAR. In all patients, digital subtraction angiography (DSA) was used as roadmap to deploy the devices. After DSA, manual correction was performed to correct fusion overlay to match the lowest renal artery between image fusion and DSA. Registration accuracy was determined by ostium displacement (in millimeters) of the lowest renal artery (proximal accuracy) and ostium displacement of the right and left internal iliac arteries (distal accuracy), when compared to the intra-operative DSA images (see Figure 1A & 1F). Proximal accuracy was measured before and after DSA correction. Displacement accuracy was defined as follows; accurate (0-1 mm), medium (1-4 mm) and poor (>4 mm). Tips are to register with vertebral L1/L2 centered and to correct navigation markers in axial CT-view to prevent misplacement due to ostia calcification, as displayed in Figure 1 G-H. Results: For the 11 patients the mean proximal accuracy was 0.7 (0.4-0.9) mm and distal accuracy was 5.8 (1.3-12.3) mm compared to the DSA. Before DSA correction proximal accuracy was 7.4 (1.4-11) mm. With phantom data, proximal accuracy was 0.8 (0.5-1.1) mm and distal accuracy was 2.3 (0.6-1.2) mm for strategy 2. Strategy 3 resulted in a proximal accuracy of 2.6 (1.9-3.4) mm and distal accuracy of 14.0 (13.0-15.0) mm. Strategy 4 resulted in a proximal accuracy of 1.6 (1.5-1.8) mm and distal accuracy of 4.0 (2.0- 5.9) mm. See Table 1 for an overview. Conclusion: Based on this data, image fusion proximal accuracy is equal with 2D-3D and 3D-3D registration before sheath and guidewire insertion. Manual DSA correction for 2D-3D registration is required to improve accuracy. After the insertion of guidewires, the accuracy of 3D-3D registration is superior to 2D-3D registration.