Extracting neuronal functional network dynamics via adaptive Granger causality analysis.

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Network Science

Background:

• Analyzing complex systems requires understanding functional relations between network nodes from local data.
• Existing time series methods often yield static estimates, assume stationary Gaussian data, or ignore network sparsity.
• These limitations hinder precise characterization of dynamic neuronal networks crucial for adaptive behavior.

Purpose of the Study:

• To develop a dynamic estimation and inference paradigm for extracting Granger-causal neuronal network dynamics.
• To address limitations of existing methods when applied to non-stationary, sparse neuronal data.
• To enable high-resolution analysis of functional neuronal network structures.

Main Methods:

• Integration of adaptive filtering, compressed sensing, point process theory, and high-dimensional statistics.
• Development of a novel dynamic estimation and inference paradigm.
• Application and validation using synthetic and real neuronal data.

Main Results:

• The proposed paradigm effectively extracts dynamic Granger-causal network properties from complex neuronal activity.
• Analysis of mouse auditory cortex Ca2+ imaging data revealed unique functional network structures during spontaneous activity.
• Simultaneous recordings from ferret auditory and prefrontal cortex showed rapid top-down and bottom-up functional dynamics linked to attentive behavior.

Conclusions:

• The developed method provides unprecedented spatiotemporal resolution for analyzing dynamic functional neuronal networks.
• This approach offers a powerful tool for understanding the neural basis of adaptive behaviors.
• The findings highlight the dynamic interplay between brain regions during attentive behavior.