BrainNetClass

BrainNetClass constructs and classifies functional brain networks to model high-order interactions and enable connectome-based analyses for individualized disease diagnosis and neuroscience research.

Key Features:

• Advanced network construction methods: Implements methods to capture complex and high-order interactions among brain regions, extending beyond traditional static pairwise Pearson's correlation-based functional networks.
• Classification framework: Applies a rigorous classification framework based on brain connectome features to support individualized disease diagnosis.
• Comprehensive classification outputs: Produces a range of classification-related results including network representations and identification of contributing features to assess model robustness and generalizability.
• Validation on resting-state functional MRI: Demonstrated effectiveness using real resting-state functional MRI datasets.
• MATLAB implementation: Provided as a MATLAB-based platform for implementation and analysis.

Scientific Applications:

• Connectome-based disease diagnosis: Supports individualized diagnosis by leveraging connectome-derived features for classification of disease states.
• Network-level investigation of brain function and disorders: Enables study of complex and high-order interactions in functional brain networks to investigate brain function and pathology.
• Evaluation of diagnostic model robustness: Facilitates assessment of model robustness and generalizability using network outputs and contributing feature analyses.

Methodology:

Implemented in MATLAB; constructs functional brain networks using high-order interaction methods beyond pairwise Pearson's correlation; applies classification based on brain connectome features; outputs network representations and contributing features; validated on resting-state functional MRI datasets.