Normalisation of neonatal brain network measures using stochastic approaches.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Graph Theory

Background:

• Diffusion tensor imaging (DTI) and tractography are key for brain connectivity analysis.
• Current brain parcellation methods lack a gold standard, leading to variable graph structures.
• Comparing brain networks with different node numbers and correspondences requires normalization.

Purpose of the Study:

• To propose and validate methods for normalizing brain network measures using random graphs.
• To address the challenge of comparing brain graphs with varying parcellations.
• To assess the utility of normalized measures in tracking developmental changes in brain connectivity.

Main Methods:

• Utilized random graphs for normalizing brain network measures derived from DTI tractography.
• Assessed the local stability of normalized measures across different random parcellations.
• Applied the normalization method to a serial diffusion MRI dataset of neonates.

Main Results:

• Normalized brain network measures demonstrated local stability across distinct random parcellations.
• The method successfully characterized changes in brain connectivity during early development in a neonatal cohort.
• The proposed normalization scheme facilitates meaningful intra- and inter-subject comparisons.

Conclusions:

• The proposed random graph-based normalization method provides a robust approach for comparing brain networks.
• This technique enhances the reliability of brain connectivity analysis, particularly in developmental studies.
• The findings support the use of normalized network measures for exploring brain development and connectivity.