Pipelines

The goal of fMRIDenoise is to perform denoising on fMRI data preprocessed via fMRIPrep using common denoising strategies. Denoising refers to a procedure of minimizing confounding effects of non-neuronal signals (related to head motion, scanner noise, or physiological fluctuations) by regressing them out from the fMRI data.

The neuroimaging community proposed various strategies for denoising the fMRI data [Parkes2018]. Each strategy offers a different compromise between how much of the non-neuronal fluctuations are effectively removed, and how much of neuronal fluctuations are damaged in the process.

As there is currently no consensus in the fMRI community on an optimal denoising strategy that perform best on a broad range of datasets, fMRIDenoise offers a simple way to denoise your fMRI data using different denoising strategies, inspect quality measures of your denoised data for each strategy, and select the best performing one.

Confounding variables calculated in fMRIPrep are stored separately for each subject, session and run in TSV (tab-separated value) files - one column for each confound variable (read more about confounds output in fMRIPrep documentation).

Default denoising pipelines

By default, fMRIDenoise performs denoising using 6 common denoising pipelines (+ one Null pipeline with only filtering applied that can be use as a refernce). Earch pipeline is defined as a single .json file in pipelines folder. All default pipelines are described below.

24HMP8PhysSpikeReg

Denoising strategy based on regressing out: 24HMP - 24 head motion parameters including: 3 translations, 3 rotations, their temporal derivatives, and their quadratic terms [Satterthwaite2013], 8Phys - mean physiological signals from white matter (WM) and cerebrospinal fluid (CSF), their temporal derivatives, and quadratic terms [Satterthwaite2013], S pikeReg - spike regressors based on FD and DVARS thresholds [Power2012].

24HMP8PhysSpikeReg4GS

Denoising strategy based on regressing out: 24HMP - 24 head motion parameters including: 3 translations, 3 rotations, their temporal derivatives, and their quadratic terms [Satterthwaite2013], 8Phys - mean physiological signals from white matter (WM) and cerebrospinal fluid (CSF), their temporal derivatives, and quadratic terms [Satterthwaite2013], SpikeReg - spike regressors based on FD and DVARS thresholds [Power2012]. Pipeline additionally includes global signal regression (GS), its temporal derivative, and quadratic terms (4GS).

24HMPaCompCorSpikeReg

Denoising strategy based on regressing out: 24HMP - 24 head motion parameters including: 3 translations, 3 rotations, their temporal derivatives, and their quadratic terms [Satterthwaite2013], aCompCor - signals extracted from 10 orthogonal principal components (PCs) obtained separately from the eroded white matter (WM; 5 PCs) and cerebrospinal fluid (CSF; 5 PCs) masks [Muschelli2014], SpikeReg - spike regressors based on FD and DVARS thresholds [Power2012]. This denoising pipeline is complementary to the pipeline used in Functional Connectivity Toolbox (CONN, [WhitfieldGabrieli2012]).

24HMPaCompCorSpikeReg4GS

Denoising strategy based on regressing out: 24HMP - 24 head motion parameters including: 3 translations, 3 rotations, their temporal derivatives, and their quadratic terms [Satterthwaite2013], aCompCor - signals extracted from 10 orthogonal principal components (PCs) obtained separately from the eroded white matter (WM; 5 PCs) and cerebrospinal fluid (CSF; 5 PCs) masks [Muschelli2014], SpikeReg - spike regressors based on FD and DVARS thresholds [Power2012]. T his denoising pipeline is complementary to the pipeline used in Functional Connectivity Toolbox (CONN, [WhitfieldGabrieli2012]). Pipeline additionally includes global signal regression (GS), its temporal derivative, and quadratic terms (4GS).

ICAAROMA8Phys

Denoising strategy based on ICA-AROMA - method that automatically identifies and removes motion artifacts from fMRI data [Prium2015]. Pipeline additionally regress out 8Phys - mean physiological signals from white matter (WM) and cerebrospinal fluid (CSF), their quadratic terms [Satterthwaite2013].

ICAAROMA8Phys4GS

Denoising strategy based on ICA-AROMA - method that automatically identifies and removes motion artifacts from fMRI data [Prium2015]. Pipeline additionally regress out 8Phys - mean physiological signals from white matter (WM) and cerebrospinal fluid (CSF), their quadratic terms [Satterthwaite2013]. Pipeline additionally includes global signal regression (GS), its temporal derivative, and quadratic terms (4GS).

Null

Reference pipeline with no denoising strategy applied.

Adding a custom denoising strategy

You can easily add a custom pipeline by adding a .json file to the pipelines folder of fMRIDenoise. A file should follow the structure below.

Template:

{
  "name": "PipelineName",
  "description": "Pipeline description",
  "confounds": {
    "white_matter": {
      "raw": "False",
      "derivative1": "False",
      "power2": "False",
      "derivative1_power2": "False"
      },
    "csf": {
      "raw": "False",
      "derivative1": "False",
      "power2": "False",
      "derivative1_power2":  "False"
      },
    "global_signal": {
      "raw": "False",
      "derivative1": "False",
      "power2": "False",
      "derivative1_power2": "False"
      },
    "motion": {
      "raw": "False",
      "derivative1": "False",
      "power2": "False",
      "derivative1_power2": "False"
      },
    "acompcor": "False"
  },
  "aroma": "False",
  "spikes": "False"
}

References

Muschelli2014(1,2)

Muschelli J, Nebel MB, Caffo BS, Barber AD, Pekar JJ, Mostofsky SH, Reduction of motion-related artifacts in resting state fMRI using aCompCor. NeuroImage. 2014. doi:10.1016/j.neuroimage.2014.03.028

Prium2015(1,2)

Pruim RHR, Mennes M, van Rooij D, Llera A, Buitelaar JK, Beckmann CF. ICA-AROMA: A robust ICA-based strategy for removing motion artifacts from fMRI data. Neuroimage. 2015 May 15;112:267–77. doi:10.1016/j.neuroimage.2015.02.064.

Parkes2018

Parkes L, Fulcher B, Yücel M, Fornito A, An evaluation of the efficacy, reliability, and sensitivity of motion correction strategies for resting-state functional MRI. NeuroImage. 2018. doi:10.1016/j.neuroimage.2017.12.073

Power2012(1,2,3,4)

Power JD, Barnes KA, Snyder AZ, Schlaggar BL, Petersen, SA, Spurious but systematic correlations in functional connectivity MRI networks arise from subject motion. NeuroImage. 2012. doi:10.1016/j.neuroimage.2011.10.018

Satterthwaite2013(1,2,3,4,5,6,7,8)

Satterthwaite TD, Elliott MA, Gerraty RT, Ruparel K, Loughead J, Calkins ME, Eickhoff SB, Hakonarson H, Gur RC, Gur RE, Wolf DH, An improved framework for confound regression and filtering for control of motion artifact in the preprocessing of resting-state functional connectivity data. NeuroImage. 2013. doi:10.1016/j.neuroimage.2012.08.052

WhitfieldGabrieli2012(1,2)

Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain connectivity. 2012. doi: 10.1089/brain.2012.0073