A spectral approach to analyzing slice selection in planar imaging: optimization for through-plane interpolation.

Interpolation between slices is necessary whenever reslicing a volume of data into a different coordinate frame. This may be done to view the data from different perspectives, to align data from different sessions, or to remove the effects of head movement in functional imaging studies. In this paper, issues surrounding slice-selection in two-dimensional imaging are examined in the context of through-plane interpolation and a spectral framework is introduced to describe the sources of error when interpolating between slices. This framework suggests that there is a trade-off between precision in localization, which requires high spatial frequencies, and interpolation, which requires a narrow spectrum of spatial frequencies. An analysis of the sources of error has lead to several approaches to reducing interpolation error including elimination of interslice gaps or making slices overlap, use of a slice profile with a narrower spatial frequency bandwidth such as a Gaussian profile, and use of high-order interpolation. The simulation and experimental data demonstrate significant reductions in interpolation error for these approaches.