Data extrapolation for truncation artifact removal

Abstract

Clinicians typically obtain high-resolution clinical MR images in an effort to avoid the truncation artifacts that often arise in Fourier transform reconstruction of limited data. A method for reducing these artifacts in MR images, at the reconstruction stage, would allow for reduced imaging times, through the collection of fewer phase encode steps and increased signal-to-noise ratios, through increased pixel size. The approach to reducing truncation artifacts in MR images is developed and a simple algorithm is presented which significantly reduces truncation artifacts in images with as few as 96 phase encode steps. The algorithm is compared with a more sophisticated method of reconstructing truncation-free images and is shown to be equivalently effective. Three clinical examples are shown illustrating the Success of the method. © 1991 Academic Press, Inc.