Convolution power spectrum analysis for FMRI data based on prior image signal.

Area of Science:

• Neuroimaging
• Signal Processing
• Computational Neuroscience

Background:

• Conventional functional MRI (fMRI) analysis often models blood-oxygen-level-dependent (BOLD) responses over time.
• Power spectrum (PS) analysis focuses on the dynamic energy changes within interacting systems.
• Existing methods like statistical parametric mapping (SPM) and support vector machines have limitations in capturing complex fMRI dynamics.

Purpose of the Study:

• To introduce a novel convolution power spectrum (CPS) analysis for detecting brain functional activation in fMRI data.
• To enhance the detection of BOLD signal changes by focusing on dynamic energy transformations.
• To provide a complementary method for analyzing complex fMRI time series.

Main Methods:

• Developed a CPS analysis based on matched filtering theory for fMRI data.
• Computed convolution signals between fMRI signals and experimental patterns to reduce noise.
• Applied PS density analysis to the convolution signal as a quantitative index of BOLD signal change.

Main Results:

• The CPS method demonstrated more effective detection of certain aspects of brain functional activation compared to canonical PS SPM and support vector machine methods.
• Simulation and in vivo fMRI studies, including block-design experiments, validated the CPS approach.
• The CPS method proved useful in revealing brain functional information within complex fMRI time series.

Conclusions:

• The proposed CPS analysis is a valuable tool for detecting brain functional activation in fMRI.
• CPS analysis offers a novel perspective by quantifying dynamic energy changes in BOLD signals.
• This method serves as an effective complement to existing fMRI analysis techniques for understanding brain function.