Statistical inference in brain graphs using threshold-free network-based statistics

Area of Science:

• Neuroimaging
• Graph Theory
• Statistical Inference

Background:

• Brain networks are increasingly modeled as graphs in neuroimaging research.
• Statistical inference for edge-wise connectivity in brain graphs faces challenges with false positives.

Purpose of the Study:

• To assess the properties of Threshold-Free Network-Based Statistics (TFNBS) using simulated data.
• To evaluate TFNBS's ability to control false-positive rates in brain graph analysis.

Main Methods:

• TFNBS combines threshold-free cluster enhancement with Network-Based Statistics (NBS).
• The study utilized simulated data to test TFNBS performance.
• Unlike NBS, TFNBS provides edge-wise significance without a predefined threshold.

Main Results:

• TFNBS can be configured to detect strong, clustered effects.
• The method demonstrates effective control over false-positive rates.
• TFNBS proved sensitive to topological effects within brain networks.

Conclusions:

• TFNBS is a suitable technique for statistical assessment of brain graphs.
• The findings support TFNBS as a valuable tool for neuroimaging studies.
• Careful parameter selection is important for optimal TFNBS performance.