Motion outliers

There are new parameters for flagging a volume as an outlier.
What are the consequences on a volume flagged as outlier in the pipeline?
Which stages use this flag (I assume it’s used in the report and in the the output tsv’s etc. any other stage is influenced?)

Hi @brai, I’m not sure I understand the statement:

Can you clarify? If you’re referring to the --dumy-scans flag, that allows you to indicate that the first n volumes are intended to be discarded, and will override any automatic detection of scans.

As to the rest, the only effect of detected non-steady-state volumes is to exclude them from CompCor and ICA-AROMA. These will be affected by --dummy-scans.

Thanks, but I wasn’t refering to the dummy scans-
according to the docs, the new thresholds (–fd-spike-threshold) etc.- causes certain volumes to be counted as “motion outliers” (fmriprep’s nomenclature- not mine- taken from some page in the docs or from the code).
I was wandering what are the consequences for these, except for maybe the report stats etc.
Many thanks again!

Oh, I see. I’d forgotten about those.

From the docs:

All these confounds can be used to perform scrubbing and censoring of outliers, in the subsequent first-level analysis when building the design matrix, and in group level analysis. Spike regressors for outlier censoring can also be generated from within fMRIPrep using the command line options --fd-spike-threshold and --dvars-spike-threshold . Spike regressors are stored in separate motion_outlier_XX columns.

I don’t believe that these have any other effect except to produce motion outlier columns for downstream processing. @rastko, is that correct?

1 Like

Yeah, that’s exactly right – the “motion outliers” (which can be computed on the basis of FD, DVARS, or a combination) are generated for potential use in downstream workflows that might do things like confound regression or scrubbing. Since fMRIPrep doesn’t perform confound regression itself, the thresholds won’t have any effect on anything that’s done by the pipeline other than generation of the spike regressor columns in the ~desc-confounds_regressors files.