Dear,
I am conducting oblique correction for EPI data in resolution of 3.125x3.125x4.50 with 3dWarp, which coming out an image in resolution of 3.125x3.125x3.125. I’m curious of whether this upsampling is alright cause it may involve more noise relative to downsampling? am i right?
To put it a general word, are there any principles we can rest on when we need to conduct resampling? Are there any limitations?
Best wishes,
Peng
Hi Peng,
If you are eventually warping the data to standard space, then align_epi_anat.py and afni_proc.py will handle the obliquity for you, so the extra resample (blur) you are doing is probably not needed.
In afni_proc.py, voxels are resampled to be isotropic at the minimum voxel dimension, truncated to 3 significant bits. Using the minimum dimension means you are not losing resolution after any rotation or shift. Oversampling any larger dimensions will always come with that.
Oversampling more than this does not seem so appropriate, along the lines of correction for multiple comparisons and for accurately representing the data to readers of the papers. Permutation and FDR based corrections should not be too affected by oversampling though. When using cluster correction, it depends on how much blur there is. If the blur is high relative to the voxel size, then oversampling will not affect that too much either.
Anyway, the default with afni_proc.py is to stay close to the original resolution, without losing it due to motion and warps to standard space.
In your case, 3.125 would be truncated down to 3.
The National Institute of Mental Health (NIMH) is part of the National Institutes of Health (NIH), a component of the U.S. Department of Health and Human Services.
Freedom of Information Act |
HHS Vulnerability Disclosure
National Institute of Mental
Health | National Institutes of Health | U.S.
Department of Health and Human Services |