Brain without anatomy: construction and comparison of fully network-driven structural MRI connectomes

Area of Science:

• Neuroimaging
• Network Science
• Developmental Neuroscience

Background:

• Magnetic Resonance Imaging (MRI) connectomics models the brain as a network, offering insights into its organization and disruptions.
• Current methods often rely on standardized anatomical atlases (e.g., Brodmann areas) for defining network nodes, which is limited by inter-subject variability, especially during brain development or neuroplasticity.

Purpose of the Study:

• To develop and validate a novel, atlas-free approach for constructing and comparing diffusion MRI-based brain networks.
• To enable connection-wise comparisons of brain networks across different age groups and anatomical variations.

Main Methods:

• Combined image processing and network theory to create an atlas-free framework for diffusion MRI connectomics.
• Employed a data-driven method to determine the optimal number of equal-area nodes, assuming complete cortical connectivity.
• Utilized matrix alignment with simulated annealing for network domain alignment to a reference brain within age groups.

Main Results:

• The novel approach demonstrated successful application in neonates, 6-month-old infants, and adults.
• Pair-wise network alignment achieved correlation coefficients ranging from 0.6102 to 0.6673.
• Reproducibility tests showed high correlation coefficients (0.7443 for 6-month-olds, 0.7037 for adults), significantly exceeding inter-subject variability.

Conclusions:

• The developed framework allows for network-driven analysis of structural MRI connectomes, abstracting from specific anatomy.
• This method is applicable to subjects at any developmental stage and with significant anatomical differences.
• The atlas-free approach overcomes limitations of traditional methods, enhancing the study of brain maturation and neuroplasticity.