Despiking parameters

[araikes]

Quick question: for the 36p+despike, the design file has the following:

confound2_framewise[2]=fds:0.167,dv:2

I have two questions:

1. A lot of the examples on readthedocs also include the rmss: field for this line. Is it necessary or is it simply implied by the fds field and TR in the current versions?

2. What is the dv parameter and do I need to edit that? Is it just the number of frames over which to interpolate?

Thanks

[a3sha2]

Yes @araikes,
we removed rmss to avoid confusion with fds. Both are the same by defination. dv is for is short form of D-VARS ( D referring to temporal derivative of timecourses, VARS referring to rms variance over voxels). You can look at this paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3254728/

[araikes]

Hi @a3sha2,
Thanks for clearing up point #1. I assumed that was the case.

For point #2, thanks for identifying that it was the dvars. I should've assumed that. In relationship to the despiking design file, does this value need to change as a function of TR? I ask because the example values in the design file are based on a TR of 3s (as identified in https://xcpengine.readthedocs.io/modules/regress.html#censoring) but I have a TR of 2s for my data and I want to make sure that I'm setting the parameters correctly.

Thanks

[a3sha2]

DVARS measures how much image intensity has changed from one frame to the next frame but FD measures relative displacement from one frame to the next. the standardized dvars usually less than 2 and a higher value of dvars usually cause by the movement (fd), so FD and DVARS are correlated to some extent (usually >0.4). There is no clear choice of dv threshold, it is sometime very important to look at the fmriprep carpet plot. for example, maximum dv here is 1.17 for 3s TR

For 2s TR, I will suggest 2 or 1.5 dv

[araikes]

Thanks @a3sha2.

After looking through my fMRIPrep outputs, 1.5 or 2 seemed to make sense.

I appreciate the input.