The registration of 2D/3D images in robot-assisted cervical disc replacement surgery

Robot assisting artificial cervical disc replacement surgery is a surgery which is operated widely, however, because the registration of 2D/3D during the operation is the key technology and very difficult, the success rate of this surgery is very low. In this paper, a 2D/3D registration method of 2D/3D aiming at this surgery based on grayscale of image is developed. In this registration method, ray casting method is used to generate DRR, normalized correlation is used as similarity measure and Powell's optimizer is used as the optimization method. A test of registration using DRR simulation aiming at this registration method is also performed, and the experimental results show that the axial displacement error is below 0.5mm and the axial rotation error is below 0.2°. Because this registration method needs only two C-arm images, it can be applied to minimally invasive surgery.