Improving delay and strength maps derived from resting-state fMRI using PCA-based denoising and group data from the...

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Data Science

Background:

• Resting-state functional MRI (rs-fMRI) relies on low-frequency signal correlations to map neuronal connectivity.
• Non-neuronal signals within rs-fMRI data are confounds but contain valuable diagnostic information.
• Previous work developed the RIPTiDe method to extract blood arrival time delay and signal strength from BOLD data for clinical insights.

Purpose of the Study:

• To develop and validate a principal component analysis (PCA)-based denoising method for rs-fMRI derived delay and strength maps.
• To enhance the signal-to-noise ratio (SNR) of these maps without prior knowledge of noise levels.
• To improve the reliability and diagnostic utility of rs-fMRI derived vascular and perfusion metrics.

Main Methods:

• Applied PCA to denoise delay and strength maps derived from Human Connectome Project (HCP) rs-fMRI data.
• Utilized spectral analysis to identify noise components.
• Implemented an optimized method for selecting PCA components based on intraclass correlation coefficients (ICC) to maximize SNR.
• Assessed signal reliability using voxelwise ICC calculations before and after denoising.

Main Results:

• The optimized PCA denoising method significantly improved signal reliability.
• Average ICC values for delay and strength maps increased by 250% and 108%, respectively.
• The method effectively separates neuronal signal from non-neuronal noise without prior noise estimation.

Conclusions:

• PCA-based denoising is an effective strategy for enhancing the quality of rs-fMRI derived delay and strength maps.
• This optimized approach improves the SNR and reliability of perfusion and blood flow metrics.
• The enhanced maps hold potential for improved clinical diagnosis and monitoring of neurological conditions like stroke.