Parallel imaging reconstructions are now ubiquitous in MR imaging for robustly accelerating image acquisitions. These techniques make it possible to expand the scope of MR imaging in both static and dynamic imaging, where the inherently long acquisition times in MRI are problematic. The basic premise behind all these reconstructions is that the data are undersampled in some form, and the missing information is estimated using a parallel imaging reconstruction that utilizes information from multiple receiver coils.

A large number of parallel imaging techniques have been described in the literature for both Cartesian and non-Cartesian acquisitions. It is beyond the scope of this discussion to comprehensively cover all techniques. Rather, the focus of a majority of this discussion will be to explain and understand two major widely utilized representative Cartesian parallel imaging techniques, namely SENSE and GRAPPA. We will explore the differences between these methods to provide a framework to understand the technology in general. We will also explore the tradeoffs created when using these techniques, between acceleration and image artifacts.