Related Experiment Video
Updated: Jul 2, 2025
12:41
Acquisition of Resting-State Functional Magnetic Resonance Imaging Data in the Rat
Published on: August 28, 2021
4.3K
Comparing data-driven physiological denoising approaches for resting-state fMRI: implications for the study of aging
Ali M Golestani1,2, J Jean Chen3,4,5
1Department of Physics and Astronomy, University of Calgary, Calgary, AB, Canada.
Frontiers in Neuroscience
|February 21, 2024
Summary
Different denoising methods impact resting-state fMRI data quality and age-related functional connectivity. ICA-AROMA and global signal regression (GSR) remove more noise but also low-frequency signals, affecting aging studies.
Area of Science:
- Neuroimaging
- Functional Magnetic Resonance Imaging (fMRI)
- Signal Processing
Background:
- Physiological noise from cardiac and respiratory signals significantly degrades resting-state fMRI data quality.
- Data-driven denoising methods are crucial for removing physiological noise when these signals are not recorded.
- High temporal resolution fMRI data is essential for accurately capturing physiological signals and evaluating denoising efficacy.
Purpose of the Study:
- To evaluate the effectiveness of various data-driven denoising methods on resting-state fMRI data.
- To assess the impact of denoising on low-frequency signal power, physiological noise, and age-related functional connectivity (fcMRI) differences.
- To investigate how denoising method performance varies across different age groups.
Main Methods:
- Utilized high-temporal resolution fMRI data to assess denoising techniques including ICA-AROMA, global signal regression (GSR), white matter/cerebrospinal fluid regression (WM-CSF), anatomical CompCor (aCompCor), and temporal CompCor (tCompCor).
- Analyzed the effects of these methods on low-frequency signal power, cardiac and respiratory noise, and age-related fcMRI.
- Validated findings using data downsampled to conventional fMRI temporal resolutions.
Main Results:
- ICA-AROMA and GSR effectively removed the most physiological noise but also reduced low-frequency signals, leading to diminished age-related fcMRI differences.
- aCompCor and tCompCor were more effective at removing high-frequency physiological signals but retained more low-frequency power, showing higher age-related fcMRI differences.
- Denoising method performance varied depending on the age group, and results were consistent across different temporal resolutions.
Conclusions:
- Direct comparison of fcMRI findings across different denoising methods in aging research should be approached with caution.
- This study enhances the understanding of how various denoising strategies influence fcMRI analysis, particularly concerning physiological noise and aging effects.
- The choice of denoising method can significantly impact the detection of age-related changes in brain connectivity.
