A wavelet method for modeling and despiking motion artifacts from resting-state fMRI time series

Area of Science:

• Neuroimaging
• Signal Processing
• Biomedical Engineering

Background:

• In-scanner head movement significantly impacts functional magnetic resonance imaging (fMRI) signals.
• Traditional motion correction methods, like linear regression, are often insufficient to remove motion confounds.
• Even minor movements can introduce spurious biases in functional connectivity estimates.

Purpose of the Study:

• To introduce a novel data-driven, wavelet-based method for identifying, modeling, and removing motion artifacts in fMRI.
• To demonstrate the efficacy of the Wavelet Despike method in mitigating motion-related biases without data scrubbing.

Main Methods:

• A spatially-adaptive, wavelet-based approach called Wavelet Despike was developed.
• This method integrates as an additional step into standard fMRI pre-processing pipelines.
• The method was validated on resting-state fMRI data from three cohorts, including high-motion datasets.

Main Results:

• The Wavelet Despike method robustly removed various motion artifacts and motion-related biases, including distance-dependent connectivity artifacts.
• The method demonstrated effectiveness at both group and single-subject levels.
• It successfully accommodated spatial and temporal heterogeneity of motion artifacts, removing high and low-frequency artifacts.

Conclusions:

• A significant risk of motion-related bias exists in fMRI connectivity analysis.
• Effective time series denoising strategies, such as Wavelet Despike, can manage this risk.
• The Wavelet Despiking software is freely available for use.