Dynamic Co-Modulation (DyCoM): A Unified Operator Framework for Dynamic Connectivity in Neuroimaging.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Dynamic connectivity analysis is crucial for understanding time-varying brain interactions.
• Current methods are fragmented, causing inconsistent findings and limited comparability.
• Attributing observed effects to specific computational choices remains challenging.

Purpose of the Study:

• Introduce dynamic co-modulation (DyCoM), a unified operator-level framework for dynamic connectivity estimation.
• Disentangle previously conflated findings by revealing distinct operator choices within DyCoM.
• Provide a principled, domain-agnostic framework for coherence, interpretability, and estimator development.

Main Methods:

• Developed DyCoM, a compact framework expressing dynamic connectivity estimators as compositions of fundamental signal processing operations.
• Utilized simulations and resting-state fMRI data to validate the DyCoM framework.
• Analyzed neurobiological signatures, state-transition sensitivity, and clinical associations within the DyCoM framework.

Main Results:

• DyCoM successfully disentangled findings previously conflated by different estimator designs.
• Demonstrated that distinct operator choices within DyCoM correspond to specific neurobiological signatures (sensory, executive control), state-transition sensitivity, and clinical associations.
• Showcased the framework's ability to reveal how estimator design choices influence biological interpretations.

Conclusions:

• DyCoM establishes a unifying foundation for dynamic interaction analysis in neuroscience.
• Highlights the critical impact of estimator design choices on biological interpretations of dynamic brain connectivity.
• Offers a principled and domain-agnostic approach for developing and interpreting dynamic connectivity estimators.