Connection density affects the behavior of functional brain network metrics

Area of Science:

• Neuroscience
• Network Science
• Computational Psychiatry

Background:

• Neuropsychiatric disorders are linked to altered functional brain network (FBN) connectivity.
• Functional network metrics quantify changes in network segregation and integration.
• Inconsistent findings in FBN studies are often due to arbitrary thresholding or binarization methods.

Purpose of the Study:

• To investigate the impact of connection density on FBN metrics.
• To explain the conflicting conclusions in existing literature.
• To establish standardized methods for FBN analysis in neuropsychiatric disorders.

Main Methods:

• Analyzed 16 functional network metrics across three datasets (Alzheimer's disease, mild cognitive impairment, schizophrenia).
• Examined connection densities from 1% to 99% in binary and weighted networks.
• Utilized time and wavelet domains for comprehensive analysis.

Main Results:

• Discovered a 'reversal phenomenon' where group differences invert with increasing connection density.
• Found that significant metrics vary by analytical mode (domain, network type) and disease.
• Identified disease-specific metric differences, highlighting heterogeneity in neuropsychiatric disorders.

Conclusions:

• Established optimal connection density ranges to mitigate the 'reversal phenomenon' and enhance inter-group differences.
• Identified robust metrics for reliable FBN analyses across datasets.
• Emphasized the critical need for standardized connection density selection in FBN research for consistent and comparable results.