Bayesian inference of structural brain networks.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Biophysics

Background:

• Diffusion-weighted magnetic resonance imaging (dMRI) and probabilistic tractography are used to model white-matter connectivity.
• Current methods for inferring brain networks from tractography data lack a principled approach, relying on arbitrary thresholds or difficult-to-interpret weighted results.

Purpose of the Study:

• To develop a generative model that explicitly links structural brain networks to observed streamline distributions from tractography.
• To enable principled inference of brain networks and validate the model using functional MRI data.

Main Methods:

• A generative model was developed to describe the relationship between structural brain networks and dMRI streamline distributions.
• The model's inferences were validated using simultaneously acquired resting-state functional MRI data.
• A prior combining multi-subject connectivity estimates was used to further inform network inference.

Main Results:

• The proposed generative model provides a principled framework for inferring structural brain networks.
• Validation with resting-state fMRI data demonstrated the model's efficacy.
• Incorporating a multi-subject prior significantly improved network inference compared to conventional methods.

Conclusions:

• The developed generative model offers a robust and interpretable method for structural brain network inference.
• This approach advances the field of neuroimaging by providing a more reliable way to map brain connectivity.
• The findings have implications for understanding brain structure-function relationships and neurological disorders.