Estimation of the Hemodynamic Response Function in event-related functional MRI: directed acyclic graphs for a...

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biophysics

Background:

• Blood-Oxygen-Level-Dependent (BOLD) functional Magnetic Resonance Imaging (fMRI) is a key tool for analyzing brain activity.
• Modeling the brain as a linear system with a Hemodynamic Response Function (HRF) is a common analysis approach.
• Estimating the HRF is crucial but challenging due to the ill-conditioned nature of the problem.

Purpose of the Study:

• To present a general Bayesian model for HRF estimation.
• To translate this model into a graphical model framework for improved representation and computation.
• To enable efficient estimation of the HRF and associated model parameters.

Main Methods:

• Recalled the most general Bayesian model for HRF estimation.
• Translated the Bayesian model into a graphical model representation.
• Developed a numerical scheme to approximate the joint posterior distribution for parameter estimation.

Main Results:

• The graphical model provides a clear and efficient representation of structural and functional relationships.
• A straightforward numerical scheme was established for approximating the joint posterior distribution.
• The novel technique was successfully applied to both simulated and real fMRI data.

Conclusions:

• Graphical models offer a beneficial framework for HRF estimation in BOLD fMRI.
• This approach facilitates a more robust and efficient analysis of brain activity.
• The method holds promise for advancing BOLD fMRI data analysis techniques.