From static to dynamic 1.5T MRI-linac prototype: impact of gantry position related magnetic field variation on image fidelity

Abstract

Recently, the MRI-linac prototype at UMC Utrecht was upgraded with a ring-based gantry, where all linac components are mounted on a ring around the MR scanner. Although adaptations have been made to both linac and MR magnets in order to account for mutual magnetic interference, interference problems cannot be ruled out completely. Therefore, in this paper the impact of gantry position dependent magnetic field inhomogeneity variation on the geometrical accuracy of acquired MR images is quantified. Magnetic field maps were acquired in a large field of view for static gantry positions in shimmed and un-shimmed conditions. Reproducibility of the shim settings was assessed. From the fieldmaps, a minimum gradient strength needed to acquire images with geometric distortions of, at most, 1 mm was derived. Moreover, imaging during gantry rotation was performed for a range of imaging parameters and rotation speeds. From the measurements we conclude that images with good geometric fidelity can be obtained for all static gantry positions, provided that shimming is performed for each new gantry position. This indicates that the present prototype is suitable for static IMRT scenarios. Shim settings are highly reproducible, suggesting that shimming via look-up tables is feasible. Finally, imaging during gantry rotation may produce severely distorted images at present and is likely to require advanced compensation methods such as dynamic shimming or higher order reconstruction.