Self-calibrating through-time spiral GRAPPA for real-time CMR

Background Through-Time non-Cartesian GRAPPA, a novel parallel imaging method for non-Cartesian trajectories, has recently been shown to provide real-time, free-breathing cardiac images with temporal resolutions of less than 35 ms per frame [Seiberlich N, et al. MRM 2011 Dec;66 (6):1682-8]. The drawback to this method is the need for several fully-sampled datasets for calibration stemming from the non-Cartesian nature of the data, which leads to a longer overall scan time. By acquiring interleaved spiral datasets and combining them to form fully-sampled datasets for the generation of the GRAPPA weights, as in TGRAPPA [Breuer FA, et al. MRM 2005 Apr;53(4):9815.], there is no need for additional calibration data. However, this interleaved calibration method poses the risk of increased artifacts if the temporal footprint of the calibration data is too long. The goal of this study is to test the extent to which self-calibrating through-time spiral GRAPPA can be used for real-time free-breathing CMR.