Ripple artifact reduction using slice overlap in slice encoding for metal artifact correction

Purpose Multispectral imaging (MSI) significantly reduces metal artifacts. Yet, especially in techniques that use gradient selection, such as slice encoding for metal artifact correction (SEMAC), a residual ripple artifact may be prominent. Here, an analysis is presented of the ripple artifact and of slice overlap as an approach to reduce the artifact. Methods The ripple artifact was analyzed theoretically to clarify its cause. Slice overlap, conceptually similar to spectral bin overlap in multi‐acquisition with variable resonances image combination (MAVRIC), was achieved by reducing the selection gradient and, thus, increasing the slice profile width. Time domain simulations and phantom experiments were performed to validate the analyses and proposed solution. Results Discontinuities between slices are aggravated by signal displacement in the frequency encoding direction in areas with deviating B0. Specifically, it was demonstrated that ripple artifacts appear only where B0 varies both in‐plane and through‐plane. Simulations and phantom studies of metal implants confirmed the efficacy of slice overlap to reduce the artifact. Conclusion The ripple artifact is an important limitation of gradient selection based MSI techniques, and can be understood using the presented simulations. At a scan‐time penalty, slice overlap effectively addressed the artifact, thereby improving image quality near metal implants. Magn Reson Med 73:318–324, 2015. © 2014 Wiley Periodicals, Inc.