Parallel computation of fan-beam back-projection reconstruction algorithm in computed tomography

A method for parallelizing the image reconstruction of a fan beam computed tomography (CT) scanner and a processor array adapted for computing the back-projection reconstruction algorithm (BPR) are presented. By making use of some inherent parallelism in PBR, a modular ring architecture in multiple of eight processors is shown to be well suited for a class of CT algorithms. The parallel architecture encompasses a reconfigurable binary tree to accommodate a wide range of processor interconnections needed for different image-processing schemes, ranging from front-end filtering, calibration, and fast Fourier transform to postprocessing such as artifact removal and image compression.<>