XCP 0.14.1 doesn't use smoothed data even though smoothing was requested

Software Support
1

Summary of what happened:

I want to use xcpd 0.14.1 to compute parcel-wise connectivity matrices in order to replicate results from a former colleague. I used the command below which is a slightly adapted version of the nichart mode. I noticed that the resulting connectivity matrices are based on the unsmoothed timeseries, even though I thought, that I can overwrite these settings by setting --smoothing 4? See this passage:

By default, the nichart mode will apply the following parameters, which may be overridden by the user:

Command used (and if a helper script was used, a link to the helper script or the command generated):

xcp_d /path/to/fmriprep_25.2.4/ /path/to/xcpd_0.14.1_parcelwise/ participant \
--fs-license-file /path/to/freesurfer_license.txt \
--n_cpus 6 \
--mode nichart \
--smoothing 4 \
--dummy-scans 4 \
--skip alff reho \
--min-coverage 0.0 \
--work-dir /path/to/workdir/ \
--participant-label PIAM27JSV

Version:

XCP-D 0.14.1

Environment (Docker, Singularity / Apptainer, custom installation):

Singularity

Relevant log outputs (up to 20 lines):

The .toml files still have the correct smoothing kernel:

xcp_d.txt (2.7 KB)

Screenshots / relevant information:

But in the CITATION.md file it says:

Corresponding pair-wise functional connectivity between all regions was computed for each atlas,
which was operationalized as the Pearson’s correlation of each parcel’s unsmoothed timeseries.

CITATION.txt (3.5 KB)

And the sidecars of the mean time series also suggest, that the unsmoothed time series were used:

sub-PIAM2G55W_ses-PEZ01_task-rest_space-MNI152NLin2009cAsym_seg-4S156Parcels_stat-mean_timeseries.txt (414 Bytes)

2

Extracted time series and correlation matrices don’t use the smoothed data. The only place where the smoothed data are used is in the smoothed dense time series (desc-denoisedSmoothed_bold). Also the ALFF data will be smoothed if--smoothing is not 0 (i.e., stat-alff_desc-smooth_boldmap).

We don’t want to smooth before parcellating because parcellation is effectively a smoothing process. Smoothing before that would cause signal to bleed across parcels.