Meeting Banner
Abstract #2994

Motion Detection and Removal from Registered fMRI Data by Independent Component Analysis Method

Wu Z, Xu Y, Xu G, Li S
Medical College of Wisconsin

In fMRI studies, head motion costs serious damage to data and motion artifacts still exist even after registration. We applied independent component analysis (ICA) method to find the remaining motion component and to remove it from registered fMRI data. A motion score for each independent component was calculated by cross correlating three translational parameters and three rotational parameters obtained during registration. The component with the largest motion score was the motion component remained in the registered data. A linear combination of the rest independent components and the corresponding coefficients was performed to remove the motion component from fMRI data.