Cone-beam tomography using c-arm x-ray projections: Complete trajectories and integration of prior CT information

C-arm x-ray imaging systems can be used for the purpose of obtaining rotational cone-beam projection data. These projections can be used to reconstruct an anatomical region of interest in a wide variety of applications. Given their size, cost, and their positional flexibility, they strike a considerable advantage over conventional imaging using CT scanners. For both diagnostic and interventional applications, there is considerable demand for 3D imaging modalities, especially in the realm of computer integrated surgery. The challenges faced by volumetric c-arm imaging stems from the fact that, the ideal set of projection images that are necessary for artifact free reconstruction is not always available. These include limitation in the angular sweep of rotational acquisition, limitation in physical size of the x-ray detector, or theoretical limitations that are caused by the geometry of a particular source trajectory.; In this dissertation we make three key contributions that address these challenges. First, we characterize missing information in the 3D Radon transform space with help of a unique visualization that displays available information on the surface of unit hemispherical shells. For various scenarios in sub-optimal imaging geometries, we evaluate a quantified measure of available information. Second, we develop two source trajectories that are capable of exact volumetric reconstruction within the entire field of view. We support this claim using both a theoretical proof, and by using a generalized cone-beam algorithm on simulated data sets. Third, we develop a method that integrates prior CT information that corrects completely truncated cone-beam data of a slightly changed anatomical scenario. We validate this method using simulated data, and real CT and flat-panel c-arm data.