Harmonized connectome resampling for variance in voxel sizes

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Physics

Background:

• Diffusion MRI (dMRI) enables human brain connectivity studies.
• Understanding white matter alterations is crucial for neurological research.
• The impact of dMRI voxel resolution on connectome data is not well-characterized.

Purpose of the Study:

• To comprehensively assess the effect of dMRI voxel resolution on brain connectome graph measures.
• To determine the statistical significance of resolution-dependent differences in connectomes.
• To provide recommendations for optimal data processing for robust tractography and connectome analysis.

Main Methods:

• Utilized Human Connectome Project - Young Adult (HCP-YA) dataset with scan/rescan data from 44 subjects.
• Generated tractography and connectomes across 15 isotropic and anisotropic dMRI resolutions.
• Performed statistical comparisons (Wilcoxon Sign-Rank Test) of graph measures between adjacent resolutions.
• Investigated translational impact by resampling data to resolutions of major national studies.

Main Results:

• Nearly all pairwise comparisons between different resolutions showed statistically significant differences in graph measures (p ≤ 0.05).
• Upsampling downsampled resolutions mitigated spatial sampling effects on tractography and connectome graph measures.
• Data similarity improved with higher resolution, even after initial downsampling, suggesting resolution's critical role.

Conclusions:

• dMRI voxel resolution significantly influences derived connectome graph measures.
• Spatial sampling effects are evident across various resolutions, impacting tractography and connectome complexity.
• Resampling dMRI data to 1 mm isotropic resolution is recommended for ensuring robust tractography and connectome generation.