DySCo: A general framework for dynamic functional connectivity
Giuseppe de Alteriis1, Oliver Sherwood1, Alessandro Ciaramella2
1Institute of Psychiatry, Psychology and Neuroscience (IoPPN) King's College London, London, United Kingdom.
Plos Computational Biology
|March 7, 2025
Summary
This study introduces the Dynamic Symmetric Connectivity Matrix (DySCo) framework for analyzing dynamic functional connectivity (dFC) in brain recordings. DySCo offers efficient computation and a unified approach to understanding brain dynamics across various scales.
Area of Science:
- Neuroscience
- Computational Neuroscience
- Brain Imaging Analysis
Background:
- Characterizing brain dynamics from high-dimensional recordings is a key neuroscience challenge.
- Dynamic Functional Connectivity (dFC) analyzes time-varying interactions but lacks a unified framework and efficient algorithms.
- Existing dFC methods are often empirical, hindering interpretation and scalability for high-dimensional data and real-time applications.
Purpose of the Study:
- To introduce the Dynamic Symmetric Connectivity Matrix (DySCo) analysis framework and its associated repository.
- To provide a common theoretical foundation and computationally efficient implementation for dFC analysis.
- To enable the study of brain activity at diverse spatio-temporal scales, including voxel-level analysis.
Main Methods:
- DySCo unifies common dFC measures into a single framework, enabling analysis of spatio-temporal interaction patterns across imaging modalities.
- It offers comprehensive measures for quantifying dFC evolution, including connectivity amount, matrix similarity, and informational complexity.
- The framework leverages the Temporal Covariance EVD (TCEVD) algorithm for efficient computation in eigenvector space, significantly outperforming matrix-space algorithms.
Main Results:
- DySCo measures demonstrated sensitivity to changes in brain configurations and consistency across time and subjects.
- The TCEVD algorithm enabled computationally demanding voxel-level dFC analysis, showcasing the framework's efficiency.
- Validation on synthetic and Human Connectome Project fMRI data confirmed the framework's utility and performance.
Conclusions:
- DySCo provides a unified, computationally efficient framework for dFC analysis, advancing the study of brain dynamics.
- The framework facilitates cross-modal translation of dFC findings and detailed characterization of brain activity patterns.
- DySCo's efficiency and comprehensive measures unlock the potential of dFC for high-dimensional datasets and real-time applications.
