Xcp-d using acompcor as nuisance regressor should we still set high pass filter?

Software Support
,
1

Summary of what happened:

I am running xcp-d on my output from fmriprep with --nuisance-regressors acompcor. After my run is finished, I found in the visual report’s Functional section: “As the aCompCor regressors were generated on high-pass filtered data, the associated cosine basis regressors were included. This has the effect of high-pass filtering the data as well.”

I am not sure if we should still use highpass filtering for Butterworth bandpass filter if the cosine-basis is already doing the high pass filter for me. in other word, should i set --lower-bpf to be 0 to disable high pass filter if i decide to use acompcor as my nuisance-regressor or should i still keep it. I currently have parameter:

–lower-bpf 0.008
–upper-bpf 0.15
–nuisance-regressors acompcor

Version:

I use xcp-d singularity image version 0.10.5

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

Singularity

Screenshots / relevant information:

image
image2760×408 97.5 KB

2

Hi @selma0817 , and welcome to neurostars!

This is what I would do.

Best,
Steven

1 Like
3

Thank you for the swift reply Steven ! I am indeed thinking of making -lower-bpf to 0 but may i ask what is the equivalence of high pass threshold for acompcor’s cosine basis? By equivalence i mean does acompcor’s build-in high pass filtering has -lower-bpf at around 0.008 or is it around another number? Sorry if this question seems to be confusing, I am new to fmri analysis.

4

Hi @selma0817

The cosine filter is 128 seconds ( or 1/128 hz).

Best,
Steven

1 Like
5

great ! thank you so much !