Proceedings of the Workshop on Alternative Sensing for Robot Perception, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS),
October 2015 · Hamburg, Germany
In surgery or interventional radiology 2D/3D overlays of X-rays combined with previously recorded 3D volumes support the physician with additional visual information. Typical X-ray systems in these fields are C-arm systems. First of all, the 3D volume and the 2D images need to be registered. Afterwards, if the system moves, errors between the real world pose and the nominal pose of the system appear. One possible way to overcome this problem is an absolute calibration of the system. With this approach the projection geometry can be defined accurately enough for a whole working volume. Therefore, measurements of the end effector pose serve as input. A potential method to determine the pose of the C-arm, is a pose estimation by X-ray observations of a calibration phantom. This work evaluates how the focal spot size, the pixel size and the dose affect the noise at the pose estimation of the C-arm by X-ray observations.