Activity-State Entropy: A novel brain entropy measure based on spatial patterns of activity.

Area of Science:

• Neuroimaging
• Information Theory
• Computational Neuroscience

Background:

• Brain entropy quantifies brain activity complexity, linked to cognitive abilities.
• Shannon Entropy, from Information Theory, measures system information capacity.
• fMRI studies often use time-series entropy, assuming it reflects complex spatiotemporal patterns.

Purpose of the Study:

• Introduce a novel measure of brain entropy: Activity-State Entropy.
• Quantify brain entropy based on underlying coactivation patterns (eigenactivity states).
• Explore the relationship between eigenactivity states and overall brain entropy.

Main Methods:

• Developed Activity-State Entropy using Principal Components Analysis to identify eigenactivity states.
• Applied the measure to simulated and real resting-state fMRI data.
• Compared Activity-State Entropy with Sample Entropy and Dispersion Entropy.

Main Results:

• Activity-State Entropy effectively captures spatiotemporal complexity in simulated fMRI data.
• In resting-state fMRI, major eigenactivity states involved large coactivating voxel clusters, including Default Mode Network regions.
• Higher brain entropy correlated with greater influence from smaller, more sparsely distributed eigenactivity states.

Conclusions:

• Activity-State Entropy offers a novel way to measure the spatiotemporal complexity of brain activity.
• This new measure complements existing time-series based entropy analyses in neuroimaging.
• Findings suggest a link between specific large-scale network activity patterns and overall brain entropy.