A versatile intensity-based 3D/2D rigid registration compatible with mobile C-arm for endovascular treatment of abdominal aortic aneurysm

Purpose Augmented reality-assisted surgery requires prior registration between preoperative and intraoperative data. In the context of the endovascular aneurysm repair (EVAR) of abdominal aortic aneurysm, no satisfactory solution exists at present for clinical use, in particular in the case of use with a mobile C-arm. The difficulties stem in particular from the diversity of intraoperative images, table movements and changes of C-arm pose. Methods We propose a fast and versatile 3D/2D registration method compatible with mobile C-arm that can be easily repeated during an EVAR procedure. Applicable to both vascular and bone structures, our approach is based on an optimization by reduced exhaustive search involving a multi-resolution scheme and a decomposition of the transformation to reduce calculation time. Results Registration was performed between the preoperative CT-scan and fluoroscopic images for a group of 26 patients in order to confront our method in real conditions of use. The evaluation was completed by also performing registration between an intraoperative CBCT volume and fluoroscopic images for a group of 6 patients to compare registration results with reference transformations. The experimental results show that our approach allows obtaining accuracy of the order of 0.5 mm, a computation time of <17 s and a higher rate of success in comparison with a classical optimization method. When integrated in an augmented reality navigation system, our approach shows that it is compatible with clinical workflow. Conclusion We presented a versatile 3D/2D rigid registration applicable to all intraoperative scenes and usable to guide an EVAR procedure by augmented reality.