Cone-beam artifact reduction using second pass radon space filling

Volume reconstruction from incomplete cone-beam data is prone to artifacts because of the insufficient amount of measured data. A second pass method is proposed which reduces these artifacts by extrapolating the measured data in order to simulate complete cone-beam data. To this end a first pass reconstruction is segmented and forward projected into a virtual geometry satisfying Tuy’s completeness condition. The resulting simulated data is then reconstructed together with the measured data using an exact cone-beam reconstruction algorithm. Numerical examples are given for both measured data of a physical head phantom and for simulated data using the Forbild head phantom.