I worked out that the problem was I had also tried to resample my functional images to 1x1x1mm3 (but quickly ran out of disk space) and did not delete these normalized files before resampling them to 2x2x2mm3. I thought the normalization files were being overwritten when I called spm_write_sn() again (as I was going back to the slice-time-corrected and realigned files -- and normalization and voxel size info of the new resampled/normalized files looked right), but I checked by deleting only some of the prior 1x1x1 normalized functional scans -- and sure enough, only the undeleted/overwritten files were unexpectedly large files corresponding to the size of the prior 1x1x1 resampling while the other created-from-scratch normalized and resampled scans were about 1 MB, as would be expected.
|