Can Motion Artifacts be Completely Removed from fMRI-Activation Maps?

Abstract

Most functional magnetic resonance imaging (fMRI) studies are based on the detection of small timedependent signal variation induced by changes in blood oxygenation during brain activation. However, since this signal change is very small (typically 1-5%; even under optimal conditions and high magnetic field strengths), head motion as small as 1 mm translation or 1° rotation, can also produce spurious activation, if correlated with the stimulus paradigm. Several methods have been proposed to retrospectively (during image post-processing) or prospectively (in real time) correct for motion, however these techniques generally correct only for geometric rigid-body effects. Head motion can also change the magnetic field homogeneity, altering the effective relaxation rates of tissues (R2*), and producing timedependent geometrical distortions in fMRI studies. Unfortunately these non-linear motion-related artifacts become worse at high field strengths and cannot be corrected by standard image realignment methods; therefore, the range of motion is restricted for in vivo fMRI studies, especially at high field strengths. Real-time motion monitoring during fMRI can provide highly accurate information on whether subject motion during an fMRI scan was acceptable or excessive, and whether repeat scans with excessive motion are necessary. Keywords: fmri, brain movement, motion detector, motion artifacts, motion correction, spurious activation