Integration of higher-order dynamic fields into MR image reconstruction

Introduction Conventional magnetic resonance imaging uses linear gradient fields for spatial encoding, allowing the use of Fourier based reconstruction methods. The incorporation of dynamic non-linear phase terms into reconstruction is uncommon, as its evolution during the readout is usually unknown and reconstruction techniques have so far not been available. Recent improvements in magnetic field monitoring opened up the possibility of capturing higher-order dynamic field evolution, providing the necessary information to correct for the image distortions they cause. The present work introduces a technique for higher order field reconstruction based on monitored higher order phase evolution. The method is applied to phantom and in-vivo diffusion weighted EPI. Diffusion imaging is a particularly challenging application, since parametric maps are calculated from sets of variably diffusion weighted (DW) images. Hence the geometrical congruence between different images is of utmost importance but hampered by strong eddy current induced magnetic field perturbations during the data readout, varying with the diffusion encoding. Methods 2D single-shot spin-echo EPI data (TE=80ms, TR=5000ms, 76 phase encodes, readout duration=45.8ms, FOV=230mm) was acquired in a coronal (in-vitro) and a transverse (in-vivo) plane. Diffusion weighting (b=1000 mm/s) was applied in the frequency-encoding, phase-encoding and slice-selection direction; additionally nonDW (b0) reference data was acquired. To facilitate in-vitro data analysis a spherical phantom filled with low-diffusive silicon oil was employed. All imaging scans were performed on a 3T Philips Achieva system using an 8-channel head coil. Subsequently the imaging object and coil were removed and replaced by an array of 16 NMR probes [1] uniformly distributed on the surface of a sphere with a diameter of 20 cm. The scans were repeated, acquiring the phase data from the 16 probes which allows for (least-squares) fitting of the higher-spatial-order phase model [2] to