ALFF and temporal censoring

Dear all, thanks for the effort in creating this tool!

I have a minor question. I always assumed that ALFF could not be computed whenever you censor data (see below for two posts from afni). I see XCP_D has both ALFF and temporal censoring as default. Does XCP_D deal with this in a way that minimizes this issue, or am I missing something? I do know you can just disable temporal censoring, but I was just wondering since in my case I’d love to be able to do censoring (for QQ reasons) but also be able to calculate ALFF (ideally, also fALFF). Relatedly, would this only be an issue for fALFF since censoring would mostly compromise the calculation of frequencies outside of the bandpass range?

(see Alff/FAlff using AFNI - AFNI Message Board - AFNI Discuss Message Board and Re: censoring and 3dRSFC)

XCP-D automatically disables ALFF if censoring is enabled.

1 Like

Note that if you do have censoring in your data, you can use AFNI’s 3dLombscargle and 3dAmpToRSFC together to estimate RSFC parameters.

The Lomb-Scargle transform is essentially a generalization of the Fourier Transform, for estimating a power/magnitude spectrum from a non-uniformly sampled time series.

The spectral estimates will become increasingly biased as a large fraction of censoring occurs. Here is a poster discussing+describing that:

–pt

2 Likes