Virginia Tech
A Control-Theoretic Framework for Resting-State Network Dynamics: State-Space Modeling and Closed-Loop fMRI Experiments
Abstract
dc:description.abstractResting-state networks (RSNs), including the salience network, default mode network (DMN), and central executive network, are fundamental features of large-scale brain dynamics. However, most functional magnetic resonance imaging (fMRI) analysis relies on correlation-based methods that have limited insight into the mechanisms governing interactions among these networks. This work introduces a dynamical-systems and control-theoretic framework in which RSN activity is modeled as an input-driven system evolving over time. Linear state-space models were estimated from multivariate RSN time series, capturing intrinsic network interactions through system matrices while representing task conditions as external inputs. Controllability analysis was then applied to quantify how task inputs influence coordinated network dynamics. Results show that task-driven RSN activity exhibits structured, low-dimensional behavior, with clear separation between task conditions corresponding to transitions in cognitive state. Sensitivity and robustness analyses indicate that these dynamics are governed by a structured subset of parameters and remain stable under parameter perturbations. Model simulations reproduce the dominant, task-aligned structure of empirical RSN time series, suggesting that the estimated system captures meaningful features of large-scale brain activity. Offline adaptive simulations further demonstrate that simple state-dependent input strategies can reliably influence network dynamics across participants. Finally, a real-time adaptive implementation of the Multi-Source Interference Task (MSIT) shows that task conditions can be modulated based on ongoing neural state, producing systematic changes in both neural trajectories and behavioral measures. These findings provide evidence that RSN dynamics can be modeled, interpreted, and influenced within a unified control-theoretic framework. This establishes a foundation for closed-loop neuroimaging experiments in which brain activity is not only observed, but actively guided.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- masters
- Discipline thesis:degree_discipline
- Electrical Engineering
- Department dc:contributor.department
- Electrical Engineering
- Grantor dc:publisher
- Virginia Tech
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Dillard, Abigail Patricia
- Chairs dc:contributor.committeechair
-
- Williams, Ryan K.
- LaConte, Stephen Michael
- Committee members dc:contributor.committeemember
-
- Southward, Steve C.
- Jones, Creed Farris
Subjects
dc:subject × 5Rights
dc:rights- Statement dc:rights
-
- In Copyright
- Licence dc:rights.uri
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Other
- vt_gsexam:47208