Aggressive vs nonaggressive denoising

[mri]

I read many posts on the difference between aggressive and non-aggressive denoising using ICA AROMA, but I’m still very confused.
I thought the difference refers to only whether we use regress out nuisance variables before the full model with variables of interest (aggressive) or add nuisance regressors to the full model (nonaggr). At least that is what @mmennes seems to imply here:

Regressing out WM and CSF prior to your 1st level model (without including any variables of interest) would be classified as aggressive denoising.

However in this jupyter notebook, nonagg denoising is done by regressing out confounds, which according to @mmennes definition would be classified as aggressive denoising, wouldn’t it?

[mmennes]

if your design matrix only includes confound variables, with the idea of using the residuals as input for a next analysis, you are doing aggressive denoising.

if your design matrix includes confound variables AND variables of interest, you are doing non-aggressive denoising

[mri]

Thanks @mmennes. I think it’s clear to me what aggressive denoising is, but I’m still confused with non-aggressive denoising after AROMA.

When I run AROMA I have the option of choosing either aggressive or non-aggressive denoising (or both). The aggressive output is essentially the residuals from a nuisance regression on noise ICs, and I think the non-aggressive output is the linear regression with mixing matrix + noise ICs, is that right?

Extending this logic to CSF and WM regressors, the agressive denoising is just the residuals of the nuisance regression on the CSF and WM regressors. But how would non-aggressive denoising be implemented here? After getting the AROMA epi_nonaggressive.nii file, how would I go about doing non-aggressive denoising of CSF and WM regressors?

[mmennes]

you would add the CSF and WM regressors to a design that also holds your variables of interest, e.g., task stimulus time series for a task analysis or seed-timeseries for a seed-based functional connectivity analysis.

The point is that you either let your noise variables soak up all the variance they ‘want’ (aggressive), or that you make sure the variance they share with good variables is properly modeled (and thus not removed) (non-aggressive).

[dvsmith]

Thanks for the helpful clarifications. I had some related questions on a different thread that might haven gotten buried (ICA AROMA agg vs non-agg):

ICA AROMA agg vs non-agg

1. With the non-aggressive denoising, we are basically using a matrix that contains both signal and noise, and we are saying that any shared variance between signal and noise will not be removed. But, if there is shared variance between columns that are denoted as “noise” (e.g., perhaps a CSF regressor is highly collinear with a white regressor and the motion parameters), then the shared variance here would be removed (because it is just noise anyway)? Is that correct?
2. Assuming I am on the right track with how I’m thinking about the non-aggressive denoising, I am curious how this would play out in the analyses. For example, if we have stimulus-correlated head motion, then it would probably not be removed in the case of non-aggressive denoising? Would this then bias our stats/inferences? It seems like the motion parameters would need to be included in the analysis to prevent this from happening, but would their inclusion lead to the reintroduction of noise that was removed earlier?

[mmennes]

1. yes

2. Stimulus-correlated head motion is the worst as you can’t disentangle artefact from signal. As you indicate non-aggressive denoising would likely not (fully) remove this type of artefact (matrix algebra does not know our intentions ;-)). This would alter your stats, but you can’t know whether it’s biased or not as you don’t know which portion is stimulus-related activity and which portion is motion-induced. This is where ICA-based denoising could have a potential benefit as ICA might be able to separate these two sources. In this case the inclusion of the motion parameters is not needed anymore as the ICA components (hopefully) provide more fine-grained regressors.

[dvsmith]

Thanks, Martin! I think we’re just going to try and make one confound matrix (including the cosineXX regressors for hp filtering) during the first-level analyses in FSL. Based on all the related threads (see links below), it seems like that would be the best way of accounting for shared variance and avoid re-introducing noise.

Within that confound matrix, we would include the cosineXX regressors, the non-steady state regressors, and CSF/WM regressors alongside the aromaXX regressors. I’d like to limit loss of tDoF, but I wonder if there’s any advantage in also including the 6RP (or even the 24RP) motion parameters as an added layer of protection again head motion? I realize that some of the aromaXX regressors should be correlated with the rotations and translations (and their expansions), but I don’t think any of the evaluation papers (e.g., yours or the Lydon-Staley et al., 2019, Network Neuroscience) included a pipeline with AROMA and motion regressors.

Best,
David

Links to related threads in case it’s helpful for others:

Confounds from fmriprep: which one would you use for GLM?

Hello experts, after running fmriprep, I get a very nice tsv with the computed confounds. Would you use all of them as nuisance regressors in a GLM? Or are some more suited for resting state data vs. task-based data? Thanks! Matteo

fMRIPREP: Which regressors for best Global Signal Regression Comparison

Hi all, I ran into Chris M. at OHBM and he pointed out that when using the aCompCor regression strategy, the cosine regressors should be used too (this is not documented, we should change that) Can anyone verify that the comparison below makes sense: We are moving from bash scripts that implemented Thomas Yeo’s pipeline at MGH to fmriprep, and want to be able to compare traditional global signal regression to CompCor and/or ICA-AROMA: For the aCompCor regression: aCompCor00, aCompCor01, aCo…

fMRIPrep: nonsteady states added back into AROMA non-aggressively denoised outputs

Thanks very much! This makes sense. I have come across another issue that baffles me. I have 5 runs for one participant and at least for 2 of them, the *space-MNI152NLin2009cAsym_desc-smoothAROMAnonaggr_bold.nii file has its first three volumes non-smoothed, it looked like the nonsteady volumes are added back after ica-aroma, but not smoothed. Again, this happens for two runs out of the 5 and for several subjects. Why do you think that is?

FMRIPREP: appling ICA-AROMA and other regressors

Hi all, If I want to use ICA-AROMA to denoise my data, and use tcompcor and acompcor as additional nuisance regressors, does the order in which I apply them matter? Or do I need to find a way to apply them at the same time? To ask another way, am I in danger of reintroducing noise with another regressor after applying ICA-AROMA, since I’m applying those regressors (tcompcor, acompcor) to already partially cleaned data (ICA-AROMA denoised), while the regressors (tcompcor, acompcor) were derived …

Fmriprep with ICA-AROMA practices for PPI analysis

Hello, I was wondering if anyone had any suggestions about best practices for using fmriprep + ICA-AROMA processed data for a PPI analysis? I have already seen a few threads about using data and noise regressors generated by the fmriprep + ICA-AROMA pipeline but I haven’t found a definitive answer to my question. I think the main issue here relates to the CSF and WM regressors, which are particularly important for PPI. I understand that in fmriprep these are calculated pre-AROMA, hence using t…

Fmriprep with ica-aroma and high-pass filtering on multiband data = artifacts

Hi all, I’ve been using fmriprep with ica-aroma to analyze a multiband dataset from a 2 minute long continuous scan, and then applied a separate highpass filter. This procedure yielded an artifact, apparent when performing inter-subject correction of the full timecourses within each voxel. In the image below you can see unusual correlations in voxels that seem to include a lot of CSF, such as in sulci and ventricles (in addition to the expected higher correlations in auditory cortex). This arti…

Testing different noise models output from fmriprep - unclear results with ICA-AROMA

Greetings, I am testing the effects of various pre-processing “noise model” regressors (columns in the *_confounds.tsv from fmriprep) on rsfMRI connectivity’s association with motion (mean Framewise Displacement, in the *_confounds.tsv) on a dataset of ~325 young adults scanned on the same Prisma scanner. To do the noise model regression, I am using AFNI’s 3dTproject (in nipype), which affords simplicity in that it allows for regressing out “noise” time-series, censoring (…zeroing, or interpola…

ICA AROMA agg vs non-agg

Hi, I am aware of using the --use-aroma flag, to run ICA AROMA. However, I’m curious whether there is an option to choose to run agg vs non-aggressive ICA aroma. If not, is there a way I could use the outputs from the --use-aroma flag to perform aggressive ICA aroma? Settings: singularity, fMRI prep v1.2.5 Cheers, Thapa

fMRIprep+ICA-AROMA filtering including WM, CSF, etc. confounds in fsl_regfilt

Hello, I am using fmriprep to preprocess my rsfMRI data and I am curious if appending ‘CSF’ and ‘WM’ columns from the *confounds.tsv (as well as perhaps other columns, e.g., ‘GlobalSignal’) to the MELODIC mixing matrix is an acceptable approach for simultaneously regressing out “noise” independent components and physiological signals. I have some Nipype code for doing this ( github.com/sjburwell/fmriprep_denoising ), but the process goes as follows: run fmriprep with the AROMA option turned ‘…