Functional magnetic resonance imaging brain activation directly from k-space

Area of Science:

• Neuroimaging
• Biophysics
• Signal Processing

Background:

• Functional magnetic resonance imaging (fMRI) typically analyzes brain activation from reconstructed, magnitude-only image data.
• Current methods treat image reconstruction and statistical activation as separate processes.
• This separation may limit the full utilization of the acquired complex-valued k-space data.

Purpose of the Study:

• To develop a framework for performing statistical fMRI analysis directly on original, complex-valued k-space measurements.
• To establish a relationship between k-space measurements and reconstructed image data for direct analysis.
• To enable examination of preprocessing effects in k-space on fMRI activation and correlation.

Main Methods:

• Reviewed the relationship between complex-valued k-space measurements and reconstructed complex-valued image data.
• Expressed voxel time series measurements in terms of original spatiotemporal k-space data.
• Developed methods to determine voxelwise fMRI activation and spatiotemporal covariance in image space using k-space data.

Main Results:

• Demonstrated that fMRI activation can be determined in image space directly from original k-space measurements.
• Showed that spatiotemporal covariance of voxel time series can be derived from k-space data.
• Enabled the examination of how k-space preprocessing impacts fMRI activation and correlation.

Conclusions:

• A novel framework allows direct statistical analysis of fMRI data in k-space.
• This approach leverages the original, complex-valued measurements more effectively.
• The framework facilitates a deeper understanding of preprocessing impacts on fMRI results.