Using topological data analysis to compare inter-subject variability across resting state functional MRI brain...

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain-Behavior Research

Background:

• Resting-state functional MRI (rfMRI) data requires extensive post-processing to study brain-behavior associations.
• Deriving brain representations involves dimension reduction and feature selection, leading to variability that hinders reproducibility.
• Existing variability in brain representations complicates the integration and comparison of findings across studies.

Purpose of the Study:

• To enable direct comparison of different rfMRI brain representations.
• To investigate the impact of brain representation choice on individual differences.
• To establish best practices for assessing replicability and generalizability in rfMRI brain-behavior research.

Main Methods:

• Utilized persistent homology to analyze topologies of subject-space data.
• Evaluated 34 distinct brain representations, varying in parcellation and feature type.
• Examined how different brain representations affect measurements of inter-subject variability.

Main Results:

• Persistent homology effectively compared brain representations in the context of individual differences.
• The choice of feature type significantly influences results derived from different brain representations.
• Variability in brain representations impacts the measurement of inter-subject differences.

Conclusions:

• Feature type is a critical consideration when comparing findings across different rfMRI brain representations.
• Methodological choices in brain representation significantly affect the assessment of individual differences.
• Standardized approaches are needed to improve the replicability and generalizability of rfMRI findings.