Influence of preprocessing pipelines on lifespan network neuroscience findings
Poster No:
1592
Submission Type:
Abstract Submission
Authors:
Jason Nomi1, Aanchal Kasargod2, Danilo Bzdok3, Jingwei Li4, Taylor Bolt5, Catherine Chang6, Zachary T. Goodman7, Thomas Yeo8, Nathan Spreng3, Lucina Uddin9
Institutions:
1University of California, Los Angeles, Los Angeles, CA, 2UCLA, Los Angeles, CA, 3McGill University, Montreal, Quebec, 4Research Center Jülich, Jülich, Germany, 5UCLA Health, Los Angeles, CA, 6Vanderbilt University, Nashville, TN, 7University of Miami, Coral Gables, FL, 8Centre for Sleep and Cognition, National University of Singapore, Singapore, Singapore, 9Department of Psychology, University of California Los Angeles, Los Angeles, CA
First Author:
Co-Author(s):
Introduction:
Previous fMRI research shows that preprocessing choices significantly influence aspects of functional connectivity (FC) strength. For instance, global signal regression (GSR) can strengthen resting-state FC associations with behavior1, reduce head motion influences in FC associations2, but also skew FC correlation distributions in a negative direction3. Yet, few studies have examined how GSR and other preprocessing decisions influence lifespan age-FC associations. The current study examined how scrubbing, bandpass filtering, GSR, spatial independent component analysis (ICA), and temporal ICA may influence linear and quadratic lifespan age-FC estimates.
Methods:
A lifespan dataset (601 subjects; 6-85 years; 240 males; 10 minute resting-state scan) was downloaded from the Nathan Kline Institute data repository (Siemens Trio 3.0T; multiband (factor of 4) eyes-closed (TR = 1.4s).
The first five images were removed. Data were then despiked, realigned, normalized into 3mm MNI space, and smoothed (6mm FWHM). We evaluated six preprocessing pipelines (each with and without scrubbing and/or band pass filtering). The twenty-four pipelines were carried out on ROI-to-ROI whole-brain analyses.
Minimally Preprocessed
No covariate regression applied.
Covariate regression
Friston 24 motion parameters, white matter, cerebral spinal fluid, and a linear detrend were regressed out of the data.
ICA FIX
The ICA-FIX classifier was trained via visual-classification of noise components from 24 subjects. FMIRB's ICA-FIX classification algorithm classified noise and non-noise components from individual subject data before conducting nuisance regression of noise components. Next, the Friston 24 motion parameters and linear trends were regressed out of the data.
Temporal ICA
The temporal ICA (tICA) pipeline first exercised a group spatial ICA (sICA) producing 125 independent components (31 noise). Thirty-nine noise components were regressed out of the remaining 86 non-noise component time-courses which were then fed into a tICA producing 75 components (19 noise). Finally, the 19 noise tICA components and the 39 sICA noise components were regressed out of the data.
Global Signal Regression
The average time-series of gray-matter voxels was regressed out of the data.
Filtering
A bandpass filter of 0.01 – 0.1 Hz was applied.
Scrubbing
TRs > 0.5 mm FD were removed from analyses.
Whole-brain ROI-to-ROI Analysis
Product-moment Fisher-z transformed correlation matrices (79,800 pairs) from the Schaefer 400 ROI 17 Network Parcellation were created for each subject for each preprocessing pipeline. Regression models where FC strength was the dependent variable and age (or quadratic age) was the independent variable of interest were run with nuisance covariates of head motion and sex to determine age-FC relationships.
The first five images were removed. Data were then despiked, realigned, normalized into 3mm MNI space, and smoothed (6mm FWHM). We evaluated six preprocessing pipelines (each with and without scrubbing and/or band pass filtering). The twenty-four pipelines were carried out on ROI-to-ROI whole-brain analyses.
Minimally Preprocessed
No covariate regression applied.
Covariate regression
Friston 24 motion parameters, white matter, cerebral spinal fluid, and a linear detrend were regressed out of the data.
ICA FIX
The ICA-FIX classifier was trained via visual-classification of noise components from 24 subjects. FMIRB's ICA-FIX classification algorithm classified noise and non-noise components from individual subject data before conducting nuisance regression of noise components. Next, the Friston 24 motion parameters and linear trends were regressed out of the data.
Temporal ICA
The temporal ICA (tICA) pipeline first exercised a group spatial ICA (sICA) producing 125 independent components (31 noise). Thirty-nine noise components were regressed out of the remaining 86 non-noise component time-courses which were then fed into a tICA producing 75 components (19 noise). Finally, the 19 noise tICA components and the 39 sICA noise components were regressed out of the data.
Global Signal Regression
The average time-series of gray-matter voxels was regressed out of the data.
Filtering
A bandpass filter of 0.01 – 0.1 Hz was applied.
Scrubbing
TRs > 0.5 mm FD were removed from analyses.
Whole-brain ROI-to-ROI Analysis
Product-moment Fisher-z transformed correlation matrices (79,800 pairs) from the Schaefer 400 ROI 17 Network Parcellation were created for each subject for each preprocessing pipeline. Regression models where FC strength was the dependent variable and age (or quadratic age) was the independent variable of interest were run with nuisance covariates of head motion and sex to determine age-FC relationships.
Results:
Scrubbing had little influence on the results (scrubbed vs non-scrubbed pipelines rs > 0.94). Preprocessing pipeline and bandpass filtering had a large influence on linear age-FC effects (Fig 1) where bandpass filtering had a larger influence on the histogram distributions compared to pipelines without filtering. Within and between-network negative linear age-FC relationships for the sensorimotor and dorsal attention networks were stronger in ICA denoising pipelines. Quadratic age-FC relationships (Fig 2) were largely unaffected by preprocessing pipeline. Although bandpass filtering influenced quadratic age-FC histogram peak height, there were little age-FC differences across preprocessing pipelines.
·Figure 1: Average between- and within-network beta estimates for linear age-FC effects from the Schaefer 400 ROI parcellation.
·Figure 2: Influence of preprocessing pipeline on quadratic Age-FC associations.
Conclusions:
The results show that preprocessing pipeline choices such as bandpass filtering, GSR, and ICA denoising can influence linear age-FC relationships but have little influence on quadratic age-FC relationships. We also find that scrubbing has little influence on linear or quadratic age-FC relationships. These results demonstrate the need to carefully consider the influence of preprocessing choices on age-FC relationships.
Lifespan Development:
Early life, Adolescence, Aging 2
Modeling and Analysis Methods:
Motion Correction and Preprocessing 1
Task-Independent and Resting-State Analysis
Keywords:
Data analysis
Design and Analysis
Statistical Methods
Workflows
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Are you Internal Review Board (IRB) certified? Please note: Failure to have IRB, if applicable will lead to automatic rejection of abstract.
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
2) Power, J. D., Mitra, A., Laumann, T. O., Snyder, A. Z., Schlaggar, B. L., & Petersen, S. E. (2014). Methods to detect, characterize, and remove motion artifact in resting state fMRI. Neuroimage, 84, 320-341.
3) Murphy, K., Birn, R. M., Handwerker, D. A., Jones, T. B., & Bandettini, P. A. (2009). The impact of global signal regression on resting state correlations: are anti-correlated networks introduced?. Neuroimage, 44(3), 893-905.