Quantification of Motion-Related Artifacts in Simulated fMRI Data Using ICA

I. Drobnjak, C. F. Beckmann, M. Jenkinson Oxford Centre for FMRI of the Brain, University of Oxford, Oxford, United Kingdom, Computing Laboratory, University of Oxford, Oxford, United Kingdom Introduction: FMRI data contain many motion-related artefacts which can limit its usefulness. Many approaches exist to correct for rigid-body motion effects but they struggle to do so in cases when motion has more complex impact on the data (e.g. interaction of motion and the Bo field inhomogeneities). Independent Data Component Analysis (ICA) on FMRI data can be used to spatially and temporally identify these artefacts. However, the accuracy and usefulness of ICA for motion artefact identification and quantification needs to be tested with data which contains realistic artefacts and for which the ground truth is known. Such data can be generated using a recently developed FMRI simulator (POSSUM).We generate data that includes rigid-body motion effects for in-plane rotations, including the interactions with B0 inhomogeneities.