How spatial transformations affect functional signal

Hello community,

can someone please suggest me a reading that explains how spatial transformations (like motion correction or normalization) affect the fMRI data in temporal domain. As a part of the question I’m also curious about the affects from the fMRI data upsampling (when functional data is registered to a higher res template). Thank you!


There might be some useful coverage in the AFNI Academy series:

Specifically, this set of talks is about alignment in processing during various stages, including motion “correction” and warping to standard space:

It also comes up during the start-to-finish processing discussion:

Basically, every time you re-grid data (which happens when alignment is applied), the data gets blurred a little bit. FMRI processing includes a lot of alignment steps (motion “correction”, EPI-anatomical alignment, anatomical-template alignment, and sometimes more), so there are better and worse ways to combine these during processing. (As covered in the talks, estimating each step one at a time is fine, but then the transforms should be concatenated and applied as a single transform to the original EPI data).

And a few of these considerations are also addressed here:

Esp. in the “B” list of points, starting on page 8.


Thank you for the reply, will dig into the info!