Locating motion artifacts in parametric fMRI analysis

Abstract

In this paper we assess rigid body co-registration in terms of residual motion artifacts for the different correlation approaches used in fMRI. We summarise, from a statistical perspective, the three main approaches to parametric fMRI analysis and then present a new way of visualising motion effects in correlation analysis. This technique can be used both to select regions of relatively unambiguous activation and to verify the results of analysis. We demonstrate the usefulness of this visualisation technique on fMRI data sets suffering from motion correlated artifacts. We use it in our assesment of rigid body co-registration concluding that it is an acceptable basis for re-alignment, provided that correlation is done using a measure which estimates variance from the data at each voxel.