Extracting interpretable signatures of whole-brain dynamics through systematic comparison

Area of Science:

• Neuroscience
• Computational Biology
• Medical Imaging Analysis

Background:

• Current brain dynamics quantification relies on limited, manually selected statistical properties.
• This approach may overlook more effective dynamical features for specific applications.
• Neuroimaging studies often use resting-state functional magnetic resonance imaging (rs-fMRI) to infer brain function.

Purpose of the Study:

• To systematically compare diverse, interpretable dynamical features of brain activity from rs-fMRI data.
• To identify novel quantitative dynamical signatures for neuropsychiatric disorders.
• To develop a data-driven method for analyzing complex time-varying systems beyond neuroimaging.

Main Methods:

• Systematic comparison of intra-regional activity and inter-regional functional coupling features from rs-fMRI.
• Application of the method to case-control comparisons in four neuropsychiatric disorders.
• Evaluation of linear time-series analysis techniques for rs-fMRI data.

Main Results:

• Linear time-series analysis techniques are generally effective for rs-fMRI case-control studies.
• Simple statistical representations of fMRI dynamics performed well, particularly within single brain regions.
• Combining intra-regional properties with inter-regional coupling significantly enhanced performance, indicating distributed changes in neuropsychiatric disorders.

Conclusions:

• The study introduces a comprehensive, data-driven method for identifying and interpreting dynamical signatures in multivariate time-series data.
• The findings highlight the importance of considering both local and distributed brain activity patterns.
• The developed methodology has broad applicability to diverse scientific fields analyzing complex time-varying systems.