Josh M. Speciale1, Charles A. Bouman1,
Thomas M. Talavage1,2
1School of Electrical &
Computer Engineering, Purdue University, West Lafayette, IN, United States; 2Weldon
School of Biomedical Engineering, Purdue University, West Lafayette, IN,
United States
Parallel imaging approaches are effective at reducing scan acquisition time. However, increasing numbers of receive coils and higher acquisition acceleration factors add to the complexity of reconstructing acquired images. Our intent is to determine if complexity of image reconstruction can be reduced using sparse representations of the matrices involved without sacrificing image quality. The sparse matrix transform (SMT) approach to approximating the singular value decomposition of the SENSE unfolding matrix enables it to be represented by a series of sparse rotation matrices. This reduces the number of multiplies in the unfolding process, reducing reconstruction complexity and time.