A backpropagation filtering algorithm for cone beam tomography

A backprojection filtering algorithm is presented for cone beam tomography with circle-and-line orbit. The algorithm first converts the 2-D cone beam projections into plane integrals, which are approximated by the line integrals on the 2-D cone beam detector plane. The plane integrals are rebinned into the Radon space. The multiply measured data are properly normalized by a weighting function in the Radon domain. The frequency-domain Radon inversion formula is used to reconstruct the image. This algorithm corrects the major drawback of a previous algorithm which discarded the multiply sampled data. This algorithm uses all the untruncated plane integrals. In addition, it is flexible and can be easily tailored to any orbit