Reconstruction Algorithm for Cone-Beam X-Ray CT Using Inverse Filtering

This paper proposes a reconstruction method for the 3-D distributional image of the absorption coefficient of the object. In the method, a cone-beam X-ray source and a 2-D detector array are placed in confronting positions on the opposite sides of the object. By rotating them 360 deg., the projection data are acquired, and then the inverse filtering is applied. This is a reconstruction problem from an incomplete projection data. In this paper, the compound image is constructed by backprojecting the projection image directly along the locus of projection. The inverse filtering is applied to the result to determine the 3-D distribution image of the absorption coefficient of the object. In the defined geometrical system for the measurement, the inverse filtering cannot essentially be applied since the impulse response is space variant. Consequently, a nonuniform sampling coordinate system is defined to make the impulse response space invariant. The inverse filter function is derived from the Hankel function by utilizing the symmetry of the impulse response in regard to the axis of rotation. Furthermore, to stabilize the inverse filtering, a window function is employed which considers the space-dependence of the incompleteness of the projection image in the frequency domain. Finally, the usefulness of the proposed reconstruction method is verified by a computer simulation.