State Estimation of Hemodynamic Model for fMRI Under Confounds: SSM Method

Area of Science:

• Neuroimaging
• Biomedical Engineering
• Signal Processing

Background:

• Estimating neuronal activity from functional magnetic resonance imaging (fMRI) signals relies on hemodynamic models.
• Confounds present in fMRI signals can significantly degrade the accuracy of neuronal state change estimation.
• Existing methods struggle to effectively mitigate the impact of these confounds.

Purpose of the Study:

• To develop an advanced state-space model that incorporates confounds into conventional hemodynamic models.
• To propose a novel Confounds Square-Root Cubature Kalman Smoothing (CSCKS) algorithm for improved neuronal state estimation.
• To evaluate the performance of the CSCKS algorithm against traditional methods using simulated fMRI data.

Main Methods:

• Developed a state-space model derived from a conventional hemodynamic model, explicitly including confounds.
• Introduced the Confounds Square-Root Cubature Kalman Smoothing (CSCKS) algorithm, which re-derives key estimation parameters.
• Utilized the Balloon-Windkessel model to generate simulation data with added confounds for rigorous testing.

Main Results:

• The proposed CSCKS algorithm demonstrated a substantial reduction in estimation error.
• Under low signal-to-interference ratios (less than 21 dB), CSCKS reduced error to 16%.
• Traditional algorithms only achieved a 73% reduction in estimation error under similar conditions.

Conclusions:

• The CSCKS algorithm significantly enhances the accuracy of neuronal state estimation from fMRI data in the presence of confounds.
• This method offers a robust solution for improving the reliability of neuroimaging analyses.
• The findings highlight the importance of accounting for signal confounds in fMRI data processing.