I’m trying to find information and guidelines on how to decide which resolution to use for the template (i.e., --output-spaces). I’ve found the page in fMRIprep the explains how to specify a resolution (:res-1, :res-2, etc.) but no indications of why I should use one over the other.
For example, is it better to use a resolution most similar to the functional or the anatomical scans? Will it affect the ability to match the results with atlases that are labels (e.g., Brodmann, etc.). Thank you!
Some software, especially ML based architecture, mandate a certain size of data
You likely do not want to downsample your data, but also do not want to upsample your data too much that you’re needlessly interpolating your data and just making your file size bigger without adding information
Some operations tend to work better with isotropic voxels, which these resolutions provide
You might want to match the resolution of your T1 image to make some downstream computations/masking easier.
Always keep in mind your data are your data - id you acquire with 5mm isotropic voxels, no amount of upsampling will enable particularly meaningful insights at the sub-voxel level.