MRI Reconstruction From 2D Partial k-Space Using POCS Algorithm

In the field of magnetic resonance imaging (MRI), reconstruction from partial k-space data is a common strategy. Projection onto convex set (POCS) method has been applied to dealing with one-dimensional (ID) partial k-space. This paper applied POCS method to MR imaging from two-dimensional (2D) partial k-space data. This method is evaluated with experiments using simulate data. Compared with the zero filling imaging (ZFI) method, reconstruction results of POCS method are much better, especially by selecting partial k-space data of an optimal location. The results show that POCS method however failed to restore some missing data that has no symmetrical part in the original k-space data, leading to inevitable reconstruction errors.