Identifying Brain Network Structure for an fMRI Effective Connectivity Study Using the Least Absolute Shrinkage and...

Area of Science:

• Neuroimaging
• Systems Neuroscience
• Computational Neuroscience

Background:

• Functional magnetic resonance imaging (fMRI) is crucial for studying brain connectivity.
• Existing methods often focus on influence magnitude, neglecting network structure identification.
• Directly modeling brain networks can lead to overfitting issues.

Purpose of the Study:

• To develop a novel method for identifying brain network structures and causal influences from fMRI data.
• To address the overfitting problem in nonlinear system identification for brain networks.
• To accurately estimate both network architecture and connection strengths.

Main Methods:

• Employed a nonlinear system identification approach with a polynomial kernel.
• Applied the Least Absolute Shrinkage and Selection Operator (LASSO) model selection for network and coefficient identification.
• Validated the method on the human visual cortex using phase-encoded fMRI data and retinotopic mapping.

Main Results:

• Successfully identified network structures and associated causalities between brain regions.
• Demonstrated the ability of LASSO to select relevant connections and estimate their strengths.
• The method accurately mapped the eight-connection visual system network.

Conclusions:

• System identification combined with LASSO offers a robust solution for fMRI effective connectivity analysis.
• This approach enhances the understanding of brain network dynamics.
• It provides a powerful tool for investigating causal relationships in the brain.