Contour-based registration technique to differentiate between task-activated and head motion-induced signal variations in fMRI

Data acquired using functional magnetic resonance imaging are often contaminated by head motion. As a result, optimal information regarding task-induced (or resting-state) signal changes cannot be extracted. Intensity-based registration methods, including intensity correlation or minimum intensity variance techniques, are widely used to register two or more images. It is shown here that intensity-based registration cannot accurately register two or more images in the presence of local intensity changes arising from functional magnetic resonance, fMRI, signals. In this paper, we present a contour-based technique that can be used not only for a more robust registration, but also to help differentiate between task-induced and motion-induced signal changes. Results obtained using both phantom and human brain images demonstrate advantages of this technique compared with a conventional intensity registration technique.