Causality-based Subject and Task Fingerprints using fMRI Time-series Data.

Area of Science:

• Systems Neuroscience
• Neuroimaging Analysis
• Computational Psychiatry

Background:

• Brain network complexity necessitates advanced analytical models.
• Functional Magnetic Resonance Imaging (fMRI) is crucial for studying brain activity.
• Current methods for identifying individual brain patterns have limitations.

Purpose of the Study:

• To develop and validate a causality-based approach for fMRI fingerprinting.
• To identify unique cognitive patterns (subject fingerprints) and task-specific brain activity (task fingerprints) using causal dynamics.
• To pioneer and quantify the concept of a 'causal fingerprint' in neuroimaging.

Main Methods:

• Development of a two-timescale linear state-space model to extract spatio-temporal causal signatures from fMRI data.
• Application of modal decomposition and projection for subject identification.
• Utilization of a Graph Neural Network (GNN) model for task identification.
• Quantification of fingerprints from a cause-and-effect perspective.

Main Results:

• Demonstrated effectiveness of the causality-based approach compared to non-causality-based methods.
• Successful identification of subjects and tasks using extracted causal signatures.
• Visualization of causal signatures and discussion of their biological relevance.

Conclusions:

• The proposed causality-based fMRI fingerprinting method is effective for identifying unique individual and task-related brain patterns.
• Causal fingerprints offer a novel perspective for analyzing brain dynamics.
• Potential applications include diagnostics and monitoring in healthy individuals and neurodegenerative diseases.