 |
 |
Dear Christopher, > Dear SPM ninjas, I’m responding to a reviewer’s feedback, and they asked an astute question about the upsampling of fMRI data during normalization. I up-sampled my data from a non-isotropic voxel size (2.15x2.15x7mm) to be isotropic 2x2x2mm (the same as the target brain template). I’ve been asked to consider the effects of this on contrast statistics, and I’m unsure of them. My impression would be that upsampling the data would artificially increase the number of voxels, and therefore the number of contrasts, so that > > 1. for uncorrected p values, more between-condition significant differences will be identified. > 2. for corrected p values, the correction is calculated using the number of independent components in the data (GRF theory) which will remain constant in spite of upsampling, so that more between-condition significant differences will be identified. > > Is this correct? Am I off track? I’d be grateful for any advice. Thanks – Just to add to Stephen's comments: In the case of uncorrected p values, you are performing more tests, and so I would expect more false positives. By definition uncorrected p values do not take into account the number of tests (much less the number of independent tests), and so aren't particularly informative, anyway. I don't think that you'd run into any difficulties due to voxel size and sampling if you're correcting for multiple comparisons using random field theory. The smoothness of the data is estimated independent of the resolution; e.g. you could have a smoothness of 15 mm x 15 mm x 15 mm, regardless of whether you had 2 mm or 3 mm isotropic voxels. It's the height of a peak (relative to other peaks) and spatial extent (relative to the smoothness) that matter, as opposed to some strict relationship with the voxel resolution. So I think this is handled without a problem in SPM, but perhaps others can weigh in on the issue. Hope this helps! Best regards, Jonathan -- Dr. Jonathan Peelle Department of Neurology University of Pennsylvania 3 West Gates 3400 Spruce Street Philadelphia, PA 19104 USA http://jonathanpeelle.net/