A reconstruction algorithm for helical cone-beam SPECT

Cone-beam single photon emission computed tomography (SPECT) provides improved sensitivity for imaging small organs like the brain and heart. Current cone-beam tomography with the focal point traversing a planar orbit does not acquire sufficient data to give an accurate reconstruction. A data-acquisition method is employed which obtains complete data for cone-beam SPECT by simultaneously rotating the gamma camera and translating the patient bed, so that cone-beam projections can be obtained with the focal point traversing a helix surrounding the patient. An implementation of Grangeat's algorithm for helical cone-beam projections is developed. The algorithm requires a rebinning step to convert cone-beam data to parallel-beam data which are then reconstructed using the 3-D Radon inversion. A fast new rebinning scheme is developed which uses all of the detected data to reconstruct the image, and properly normalizes any multiply scanned data. This algorithm is shown to produce fewer artifacts than the commonly used Feldkamp algorithm when applied to either a circular planar orbit or a helical orbit acquisition. The algorithm can easily be extended to any arbitrary orbit.>