Improved dynamic parallel imaging coil calibration method robust to respiratory motion with application to first-pass contrast-enhanced myocardial perfusion imaging.

Abstract

To develop an accurate method of performing free-breathing coil calibration for application to parallel imaging reconstructions of dynamic single-shot datasets. Coil calibration data are produced through acquisition of multiple prescans before the accelerated scan, applied during free-breathing. These multiple free-breathing prescans (MFPs) provide the necessary coil information for accurate parallel imaging reconstruction of each accelerated frame of a dynamic series, under guidance of an appropriate respiratory position based matching algorithm. This is investigated in myocardial first-pass perfusion with retrospectively undersampled datasets for analysis with standard calibration techniques to guide prospectively undersampled experiments for specific demonstration of performance against a range of "temporal" calibration techniques. Reconstruction of the retrospectively subsampled datasets with MFP-calibrated parallel imaging showed significant improvements in relative root-mean-square error comparative to all other techniques (all P < 0.05; n = 6) for acceleration factors R > 3. Accelerated acquisitions, reconstructed by means of various temporal calibration techniques and analyzed by visual observer artifact scoring, also demonstrated a large improvement with use of MFPs. Artifact levels were reduced from an average of 2.5 ± 0.6 for the best performing implementation of TGRAPPA to 0.8 ± 0.4 for MFP-GRAPPA (P < 0.001; n = 20) (0 = none to 4 = strong, nondiagnostic). MFP as parallel imaging coil calibration data can give improved performance in free-breathing dynamic MR while maintaining maximal acceleration. Magn Reson Med 75:2315-2323, 2016. © 2015 Wiley Periodicals, Inc.