Fast implementations of algebraic methods for three-dimensional reconstruction from cone-beam data

摘要: The prime motivation of this work is to devise techniques that make the algebraic reconstruction technique (ART) and related methods more efficient for routine clinical use, while not compromising their accuracy. Since most computational effort ART spent projection/backprojection operations, authors first seek optimize projection algorithm. Existing algorithms are surveyed it found these either lack accuracy or speed, suitable cone-beam reconstruction. hence a new accurate extension splatting algorithm, well-known voxel-driven method. They also describe three-dimensional (3-D) ray-driven projector considerably faster than and, at same time, perfectly suited demands cone beam. then caching schemes both simultaneous (SART), which minimize number redundant computations backprojection very memory conscious. find with caching, cost an operation can be reduced equivalent 1.12 projections. SART, due its image-based volume correction scheme, harder accelerate caching. Implementations yield runtime ratios T/sub SART//T/sub ART/ between 1.5 1.15, depending on amount used.