nCREANN: Nonlinear Causal Relationship Estimation by Artificial Neural Network; Applied for Autism Connectivity Study

Area of Science:

• Neuroscience
• Computational Neuroscience
• Brain Connectivity Analysis

Background:

• Effective connectivity in the brain is crucial for understanding neural dynamics.
• Conventional methods often rely on linear models, potentially oversimplifying complex brain functions.
• Distinguishing linear from nonlinear interactions is key to a comprehensive analysis.

Purpose of the Study:

• To introduce a novel method, nonlinear Causal Relationship Estimation by Artificial Neural Network (nCREANN), for quantifying causal interactions in the brain.
• To differentiate between linear and nonlinear components of effective connectivity.
• To investigate differences in brain connectivity between children with autism spectrum disorder (ASD) and typically developing (TD) children.

Main Methods:

• Development and validation of the nCREANN algorithm using synthesized data.
• Application of nCREANN to electroencephalogram (EEG) data from ASD and TD children at rest.
• Analysis of both linear and nonlinear input-output mappings within brain networks.

Main Results:

• nCREANN successfully identified and differentiated linear and nonlinear connectivity components.
• Typically developing (TD) subjects exhibited higher overall linear connectivity.
• Autism spectrum disorder (ASD) subjects showed a more dominant nonlinear connectivity component.
• Significant differences in neural activation dynamics between ASD and TD groups were observed.

Conclusions:

• The nCREANN method offers a powerful tool for dissecting complex brain connectivity.
• Findings suggest distinct patterns of linear and nonlinear neural dynamics characterize ASD.
• This research provides new insights into brain region interactions in ASD and TD individuals.