Improved statistical efficiency of simultaneous multi-slice fMRI by reconstruction with spatially adaptive temporal...

Area of Science:

• Neuroimaging
• Magnetic Resonance Imaging
• Biomedical Engineering

Background:

• Simultaneous multi-slice (SMS) functional magnetic resonance imaging (fMRI) accelerates data acquisition but can amplify noise.
• Conventional reconstruction methods like slice-GRAPPA may suffer from g-factor noise amplification and residual aliasing.

Purpose of the Study:

• To introduce a novel temporally regularized image reconstruction approach for SMS-fMRI.
• To improve statistical efficiency and power in SMS-fMRI analyses.

Main Methods:

• Incorporation of spatially varying, encoding-dependent temporal smoothness regularization.
• Analytical derivation of efficiency gain as a function of reconstruction parameters.
• Comparison with conventional slice-GRAPPA reconstruction using simulations and experimental data.

Main Results:

• The proposed method reduces g-factor noise amplification per temporal degree of freedom.
• Demonstrated marked improvement in statistical efficiency, particularly in central brain regions.
• Confirmed that residual slice leakage and aliasing errors are not noticeably increased compared to slice-GRAPPA.

Conclusions:

• Temporally regularized reconstruction significantly enhances statistical power in SMS-fMRI.
• The approach allows for explicit parameter selection based on noise variance per degree of freedom.
• This method offers a more statistically efficient alternative for SMS-fMRI data reconstruction.