PARALLEL PROCESSING METHOD AND APPARATUS BASED ON THE ALGEBRA RECONSTRUCTION TECHNIQUE FOR RECONSTRUCTING A THREE-DIMENSIONAL COMPUTERIZED TOMOGRAPHY

摘要

the invention relates to a process and a structure of parallel processing, which is based on the algebraic reconstruction technique (art) for three dimensional reconstruct iteratively, based on the projection of a light beam tapered, a (v) of three-dimensional voxel data that represents an image of an object.it ensures the acquisition of projection data of a cone beam x-ray source using cone beam and a detector for two-dimensional matrix in order to scan the object along a scanning path for each source, a pluralit e source separated positions on (i) the sweep path sourcea set of pixels (pi) of cone beam data measured and two-dimensional. (v) all the 3d voxel data set is broken down and a plurality (m) of sub blocks of voxels (v0) and independent (vm), each containing a number of voxels.as a result of the subdivision of the voxel (v) of the three-dimensional data into sub blocks of voxels that all pixels (pi) of cone beam data and two-dimensional network (measured projection data measured is a subdivision corresponding to the ch each position in the source (i) sub set of projection data.with regions of overlap between one or more subsets of adjacent projection data corresponding to sub blocks of adjacent voxels.each sub cube voxel and a layer or sub set of projection data is processed in parallel with corresponding pairs of sub blocks of voxels and layers, or subsets of the corresponding projection data without interference.using a parallel processor architecture on two levels for the iterative reprojection and fusion and the iterative partitioning and overheads.