Compressively Sampled MR Image Reconstruction Using POCS with g-Factor as Regularization Parameter

Abstract

Compressed sensing (CS) is an effective method to reduce k-space sampling for accelerated MRI data acquisition and reconstruction. Iterative-shrinkage algorithms provide an efficient numerical technique to minimize mixed l_l − l₂ norm minimization problems. These algorithms utilize a regularization parameter to introduce sparsity in the solution for CS recovery problem. This paper introduces a new method based on geometry factor (g-Factor) as an adaptive regularization parameter. For this purpose, Projection onto Convex Sets (POCS) algorithm is modified to include regularization term in the form of g-Factor and a priori constraint (data consistency) for image reconstruction from the highly under-sampled data. The performance of the proposed algorithm is verified using simulated and actual MRI data. The results show that g-Factor as a regularization parameter provides better image reconstruction from the highly under-sampled data as compared to a fixed regularization parameter in POCS.