Analysis of interpolation effects in the reslicing of functional MR images

Typically, the final step in volume registration is the reslicing of the volume of interest. The purpose of this work is to examine the effects of this reslicing on functional MRI (fMRI) data using different interpolation methods. Functional whole-brain echo planar imaging (EPI) volumes were resliced using six different interpolation methods: trilinear, tricubic splines, and a 3D sinc function using a rectangular and a Hanning window, both with half-window lengths of 3 and 4 voxels. Interpolation by tricubic spline and 3D sinc using a Hanning window had comparable errors, although tricubic spline interpolation was computationally the fastest. Interpolation by trilinear and 3D sinc using a rectangular window had relatively large errors, although the speed of trilinear makes it desirable for some applications. Interpolations using all of the tested methods adversely affected the fMRI data, although these effects differed for each method.