Cortical parcellation based on structural connectivity: A case for generative models.

Area of Science:

• Systems neuroscience
• Neuroimaging
• Computational neuroscience

Background:

• Identifying brain networks is crucial for understanding brain function, leading to the concept of the 'connectome'.
• Connectivity-based cortex parcellation uses distinct connection profiles of brain areas for anatomical mapping.
• Current parcellation methods often incorporate prior knowledge and disregard connectivity measurement specifics.

Purpose of the Study:

• To outline the concept, prospects, and limitations of connectivity-based cortex parcellation, especially for structural connectivity.
• To advocate for connectivity-based cortex parcellation as a modeling approach for improved reliability and efficiency.
• To provide a framework for formally testing parcellation algorithm robustness and quantifying prediction precision.

Main Methods:

• Review of existing non-invasive human brain connectivity mapping techniques.
• Proposal of a modeling framework for connectivity-based cortex parcellation.
• Integration of anatomical knowledge through prior distributions and model selection for hypothesis testing.

Main Results:

• Current methods often suffer from circular argumentation and disregard anatomical specificities.
• A modeling approach allows for formal testing of parcellation algorithm robustness.
• Quantification of prediction precision and derivation of generalization statistics are possible.

Conclusions:

• Connectivity-based cortex parcellation holds promise but requires methodological refinement.
• Adopting a modeling approach enhances the reliability and efficiency of brain network mapping.
• Future frameworks should unbiasedly consider anatomical specificities and connectivity measurement apertures.