Robust ARC Parallel Imaging with 3D Prospective Motion Correction

Introduction: Three dimensional (3D) scans are susceptible to motion artifacts because scan times are often long even after acceleration with parallel imaging and because any motion affects the entire volume measurement. Prospective motion correction techniques that directly estimate rigid body motion parameters with the help of navigators and adjust scan coordinates to realign with the subject in the event of motion have been shown to successfully reduce these artifacts in volumetric acquisitions [1-3]. Use of parallel imaging is still desirable in these scans for reduction of scan time. When parallel imaging is employed in sequences that have retrospective motion correction, the parallel imaging reconstruction is performed first, followed by motion correction [4-5]. The situation is different with prospective motion correction; since the transformations of scan coordinates effectively results in a stationary object and moving coil condition (assuming a rigid coil, as is usually the case with brain imaging) during the measurement. Such data can be reconstructed using the expanded SENSE formalism proposed by Bammer et al [6]. But because of the caveats of sensitivity estimation, this work explores a data driven parallel imaging solution to this problem. Using the ARC algorithm [7], it proposes an adapted calibration+ reconstruction strategy that robustly accounts for the scan coordinate transformations affected by prospective motion correction.