Effective Brain Connectivity

Effective Brain Connectivity infers effective connectivity by integrating structural MRI (tractography/connectome) with functional MRI data to re-weight structural connectivity maps and explain observed functional co-activations.

Key Features:

• Constrained Autoregressive Model: Employs a constrained autoregressive model that re-weights an initial structural connectivity map to minimize reconstruction error while adhering to structural constraints.
• Inclusion of Indirect Connections: Distinguishes and incorporates indirect (mediated) connections to account for both direct and indirect components in functional connectivity.
• Compatibility with BOLD Signals: Applies to both raw and deconvoluted Blood Oxygen Level Dependent (BOLD) signals for fMRI analysis.
• Comparison with Dynamic Causal Models: Compares the derived dependency representation against dynamic causal models and reports closer alignment with known ground-truth data.

Scientific Applications:

• Community Detection Task: Partitions the brain into communities based on effective networks across subjects, yielding clusters that preserve structural organization while better describing functional interactions.
• Case-Control Classification Task: Aids discrimination between healthy individuals and those with autism spectrum disorder (ASD) using effective connectivity features.

Methodology:

The methodology re-weights an initial structural connectivity representation by fitting a constrained autoregressive model to functional co-activations—minimizing reconstruction error under structural constraints, incorporating indirect connections, operating on raw or deconvoluted BOLD signals, and comparing derived dependencies to dynamic causal models.