Optimal configuration for dynamic calibration of projection geometry of X-ray C-arm systems

In M. Mitschke and N. Navab, "Recovering projection geometry: how a cheap camera can outperform an expensive stereo system," Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Hilton Head, SC, USA (2000), the authors have presented a comparative study between two calibration methods for the recovery of X-ray projection geometry. The first method was based on the use of an external sensor system (Polaris from Northern Digital). The second one used a CCD camera attached to the X-ray source. It was shown that although the overall geometry is estimated more accurately, when using the external sensor system, the recovery of X-ray projection geometry based on the attached CCD camera system results in a better quality of the final 3D reconstruction. This is described by the fact that using a CCD camel a attached to the X-ray source, most of the estimation error is distributed along the main projection ads of the X-ray imaging system. In this case, the optical axes of both imaging systems are almost parallel and there is only a few centimeters distance between the X-ray source and the camera optical center The influence of errors along the main axis of X-ray imaging projection on the reconstruction is small. Motivated by this observation, the authors aimed at the best design for Camera Augmented C-arm (CAMC). The best reconstruction result would be achieved when both X-ray and optical imaging system share the exact same center of projection. This is physically impossible. However, it can be accomplished by introducing a double mirror system that bends the optical rays such that both centers of projection are virtually aligned. This results in positioning the optical camera such that it has exactly the same viewpoint as the X-ray imaging system. This results in the optimal design for Camera Augmented C-arm in order to dynamically recover the X-ray projection geometry of a moving C-arm.