A BACKPROJECTION-BASED ESTIMATION METHOD FOR REDUCING THE CONE-BEAM ARTIFACTS IN CIRCULAR TRAJECTORIES

Circular trajectories are commonly used in cone-beam (CB) CT. Insufficient data measurement in such a geometry induces CB artifacts in CT images. Development of an efficient algorithm to reduce this problem remains a major challenge. A backprojection-based method is proposed here. Each CB projection is analyzed in radon space via Grangeat's formula. Assuming the CB projection is taken from a parallel geometry, we extract those data in the unmeasured radon region. These data are then backprojected as in a parallel geometry to calculate a correction term, which is added to the FDK result with Hu's correction (H-FDK) to form a final reconstruction. Using a parallel geometry, this algorithm can be implemented efficiently on a projection-by-projection basis. The algorithm is evaluated using computer simulations, and the result shows that the proposed method greatly reduces the CB artifacts in the H-FDK reconstructions