MEPrep: A robust pipeline for multi-echo fMRI denoising and preprocessing.

Area of Science:

• Neuroimaging
• Functional Magnetic Resonance Imaging (fMRI)
• Signal Processing

Background:

• Multi-echo fMRI enhances BOLD signal quality by mitigating motion and susceptibility artifacts.
• Multi-echo independent component analysis (ME-ICA) excels at separating BOLD signals from noise compared to traditional methods.
• Existing pipelines lack advanced denoising steps for raw multi-echo data.

Purpose of the Study:

• To introduce preICA, a novel ICA-based denoising method for raw multi-echo fMRI data.
• To integrate preICA and ME-ICA into a robust preprocessing pipeline, MEPrep.
• To evaluate MEPrep's efficacy in denoising multi-echo fMRI data.

Main Methods:

• Developed preICA, an ICA-based denoising step applied before echo combination.
• Integrated preICA and ME-ICA into the fMRIPrep framework, creating the MEPrep pipeline.
• Validated MEPrep on a resting-state multi-echo fMRI dataset.

Main Results:

• MEPrep significantly improved denoising efficacy compared to existing methods.
• Key improvements included enhanced T2* model fitting, reduced motion artifacts, and increased signal-to-noise ratio.
• Functional connectivity reliability and Shannon entropy were also improved, indicating better signal preservation.

Conclusions:

• MEPrep, integrating preICA and ME-ICA, offers superior noise suppression for multi-echo fMRI.
• The pipeline enhances data quality while preserving neurobiological signal complexity.
• MEPrep provides a scalable, open-source solution for reproducible multi-echo fMRI preprocessing.