Item

Abstract

Field inhomogeneities caused by variations in magnetic susceptibility throughout the head lead to geometric distortions, mainly in the phase-encode direction of echo-planar images (EPI). The magnitude and spatial characteristics of the distortions depend on the orientation of the head in the magnetic field and will therefore vary with head movement. A new method is presented, based on a phase informed model for motion and susceptibility (PIMMS), which estimates the change in geometric distortion associated with head motion. This method fits a model of the head motion parameters and scanner hardware characteristics to EPI phase time series. The resulting maps of the model fit parameters are used to correct for susceptibility artifacts in the magnitude images. Results are shown for EPI-based fMRI time-series acquired at 3T, demonstrating that compared with conventional rigid body realignment, PIMMS removes residual variance associated with motion-related distortion effects. Furthermore, PIMMS can lead to a reduction in false negatives compared with the widely accepted approach which uses standard rigid body realignment and includes the head motion parameters in the statistical model. The PIMMS method can be used with any standard EPI sequence for which accurate phase information is available.