FAST SINC FUNCTION GRIDDING ALGORITHM FOR FOURIER INVERSION IN COMPUTER TOMOGRAPHY.

Abstract

The Fourier inversion method for reconstruction of images in computerized tomography has not been widely used owing to the perceived difficulty of interpolating from polar or other measurement grids to the Cartesian grid required for fast numerical Fourier inversion. Although the Fourier inversion method is computationally faster than the back-projection method for parallel ray projection data, the artifacts resulting from inaccurate interpolation have generally limited application of the method. The author presents a computationally efficient gridding algorithm that can be used with direct Fourier transformation to achieve arbitrarily small artifact levels. The method has potential for application to other measurement geometries such as fan-beam projections, diffraction tomography, and NMR imaging.