Robust EMG–fMRI artifact reduction for motion (FARM)

ObjectiveCurrent template-based artifact reduction methods are inadequate to reduce irregular volume- and slice-artifacts induced by limb motion in combined (surface) EMG–fMRI (electromyography–functional magnetic resonance imaging) studies. In addition, artifacts are not removed adequately for EMG frequencies above 50 Hz. We present a new fMRI artifact reduction algorithm for motion (FARM) and compare it with standard artifact correction as implemented in fMRI artifact slice-template removal (FASTR).MethodsOne control subject generated motion artifacts during EMG–fMRI. Low-frequency motion artifacts and volume-artifacts were removed prior to slice-artifact correction. Slice-artifacts were phase-shifted and removed with motion adaptive templates (FARM). EMG data were also corrected applying FASTR.ResultsTime traces demonstrate that artifacts related to sudden changes in wire position are contained to shorter time periods. EMG power spectra from neck and arm muscles show that FARM has improved performance at higher frequencies.ConclusionsHigh-pass filtering, volume-artifact removal, phase-shifting and adaptation of slice-templates to motion improve the quality of artifact-corrected EMG recorded during limb motion.SignificanceThe improved accuracy at which EMG–fMRI data can be obtained opens up new ways to directly relate self-paced movements to brain activations and to study patients suffering from movement disorders.