Differentiating BOLD and non-BOLD signals in fMRI time series using cross-cortical depth delay patterns

Area of Science:

• Neuroimaging
• Biophysics
• Signal Processing

Background:

• Functional Magnetic Resonance Imaging (fMRI) resolution has significantly improved.
• High-resolution fMRI enables novel analytical strategies for enhanced sensitivity and neuronal specificity.
• Understanding hemodynamic signal progression across cortical depths is key.

Purpose of the Study:

• To investigate the feasibility of using cross-cortical depth temporal lag (CortiLag) patterns for BOLD and non-BOLD signal categorization in fMRI.
• To develop and validate an independent component analysis (ICA)-based framework (CortiLag-ICA) for fMRI denoising.
• To offer an alternative denoising strategy for moderate- and high-spatial-resolution fMRI data.

Main Methods:

• An ICA-based framework (CortiLag-ICA) was developed, leveraging cross-cortical depth temporal dependencies.
• The framework was tested on visual task fMRI data acquired at varying spatiotemporal resolutions (1.1-2.0 mm voxels).
• CortiLag-ICA was compared to existing denoising methods, focusing on distinguishing BOLD from non-BOLD signals.

Main Results:

• The CortiLag-ICA framework successfully categorized BOLD and non-BOLD signal components.
• The method demonstrated efficacy across different spatial and temporal resolutions.
• CortiLag-ICA proved capable of identifying signal origins based on temporal lag patterns across cortical depths.

Conclusions:

• CortiLag patterns provide a viable method for differentiating neurogenic BOLD signals from nuisance factors in fMRI.
• The proposed CortiLag-ICA framework offers a promising alternative for denoising fMRI data, especially when multi-echo acquisitions are unavailable.
• This approach enhances the reliability and specificity of fMRI analyses in high-resolution neuroimaging.