Characterizing Streamline Count Invariant Graph Measures of Structural Connectomes

Area of Science:

• Neuroimaging
• Network Neuroscience
• Computational Anatomy

Background:

• Graph measures are increasingly used to analyze human connectomes.
• Lack of consensus exists on the optimal streamline count for reproducible connectome analysis.

Purpose of the Study:

• To define the relationship between streamline count and graph-measure values.
• To assess the impact of streamline count on reproducibility and repeatability of connectome metrics.

Main Methods:

• Retrospective analysis of a prospective study involving 10 healthy subjects.
• Diffusion-weighted imaging (DWI) data acquired at 3T.
• 13 graph-theory measures derived from probabilistic tractography, assessing streamline count invariance.

Main Results:

• Modularity and global efficiency showed high repeatability (ICC > 0.90) and reproducibility (CCC > 0.99) with sufficient streamlines.
• Other measures like density and characteristic path length were highly dependent on streamline count.
• Alternate definitions were provided for measures not converging with varying streamline counts.

Conclusions:

• Certain graph measures (modularity, global efficiency) are robust to streamline count variations.
• Reproducibility of other connectome metrics is contingent on using appropriate streamline counts or alternative definitions.
• Establishing standardized streamline counts is crucial for quantitative and reproducible connectome research.