A parsimonious statistical method to detect groupwise differentially expressed functional connectivity networks.

Area of Science:

• Neuroscience
• Statistics
• Graph Theory

Background:

• Group-level functional connectivity analysis is crucial for identifying differences between clinical and control groups.
• Existing methods may lack sufficient power or be prone to high false-positive rates in detecting altered connectivity.

Purpose of the Study:

• To develop a novel statistical method for detecting differentially expressed connectivity networks.
• To improve statistical power and reduce false-positive rates compared to existing approaches.

Main Methods:

• A new statistical method utilizing combinatorics graph theory and the principle of parsimony.
• Development of a network-specific test statistic and permutation testing with multiple-testing adjustment for p-value calculation.
• Validation through simulation studies and a resting-state functional magnetic resonance imaging (fMRI) case-control study.

Main Results:

• The proposed method demonstrated significantly improved power and lower false-positive rates in detecting differentially expressed connectivity networks.
• Parsimony within constrained networks effectively reduced false positives while increasing the power of informative connections.
• The new method successfully identified differentially expressed connectivity networks where existing methods were limited.

Conclusions:

• The developed statistical method offers a powerful and reliable approach for group-level functional connectivity analysis.
• This method enhances the detection of altered brain networks in clinical and experimental settings.
• The findings suggest a significant advancement in neuroimaging statistical analysis for identifying group differences.