An iterative method for fast regularized parallel MRI reconstruction
Haker S, Hoge W, Brooks D, Kilmer M, Kyriakos W
Brigham and Women's Hospital
Parallel MR imaging is an effective approach to reduce MR image acquisition time. Non-uniform subsampling allows more flexible data acquisition, and increases the potential for high quality images at high acceleration factors. Due to large problem size, iterative methods are often employed to solve the reconstruction problem. Non-uniform subsampling often requires a regularized solution due to poor conditioning, and one challenge is to choose a good regularization parameter. Typically, this requires multiple expensive system solves. Here, we present an efficient LSQR-Hybrid algorithm which provides fast reconstruction while addressing the need for rapid regularization parameter selection.