Iterative Deconvolution-Interpolation Gridding
Gabr R, Kadah Y, Aksit P, Bottomley P, Youssef A
Johns Hopkins University
We present a simple iterative solution to the problem of image reconstruction from arbitrarily-sampled k-space. The new solution solves a sparse linear system that is equivalent to deconvolution of the k-space with a small kernel. The deconvolution is accurate, and sampling onto a finer grid is not required. Avoiding grid over-sampling preserves storage crucial for 3D trajectories. The new algorithm bypasses the calculation of sampling density compensation functions, and construction of the gridding matrix is simple, fast and requires no regularization. It is thus amenable to situations involving changing k-space trajectories. The algorithm is implemented with in vivo spiral MRI.