Magnetic resonance scan‐time reduction using echo prediction

Abstract

AbstractA linear prediction (LP) filter derived from a complete echo with zero‐phase encoding amplitude is used for recovering anatomical details from a partially acquired echo sequence. The LP filter is shown to reconstruct missing k‐space phase and amplitude information, with errors sufficiently low so as to provide image reconstruction with a contrast‐to‐noise ratio (CNR) ≥ 3. For volume imaging using multislice acquisition, the partial‐echo sequence enables more number of slices to be acquired for a given repetition time period TR. For such sequences, separate predictors are used for reconstruction of missing k‐space data corresponding to each individual slice in the volume. The proposed filtering scheme is shown to achieve results comparable to other partial k‐space approaches such as singularity function analysis (SFA), when the noise content is less than about 0.4%. For higher noise levels, this technique is recommended as a preprocessing step for SFA to track the singularity locations more accurately. © 2013 Wiley Periodicals, Inc. Int J Imaging Syst Technol, 23, 1–8, 2013