The smoothing kernel is worth while to consider, but the error message points to the templates (reference = template). When using several templates or one template and a template weighting image, these images have to have the same origin / orientation / dimension / voxel size. If one of them has accidentally been reoriented the normalisation will produce exactly the same error message as reported by Shiyu.
You can replicate the error when going with some human fMRI data and selecting two templates, one of them being the SPECT.nii. For the other templates and the brain mask the origin is 92 -128 -74 with a voxel size of -2 x 2 x 2. For SPECT the origin is -92 -128 -74 with a voxel size of 2 x 2 x 2. Reslicing the SPECT onto e.g. the brain mask solves the problem.
|