[Paper Review] Temporal connection signatures of human brain networks after stroke

Catalina Obando, Charlotte Rosso+3 more|arXiv (Cornell University)|Jul 23, 2019

Functional Brain Connectivity Studies62 references4 citations

TL;DR

This study introduces temporal exponential random graph models (tERGMs) to identify dynamic brain network reorganization after stroke, revealing that the formation of intrahemispheric triangles and interhemispheric connections—particularly in the subacute phase—predict long-term functional recovery better than static network measures. These temporal connection signatures, validated on longitudinal fMRI data, serve as robust biomarkers for chronic language and visual outcomes in patients with cortical and subcortical lesions.

ABSTRACT

Plasticity after stroke is a complex phenomenon initiated by the functional reorganization of the brain, especially in the perilesional tissue. At macroscales, the reestablishment of segregation within the affected hemisphere and interhemispheric integration has been extensively documented in the reconfiguration of brain networks. However, the local connection mechanisms generating such global network changes are still largely unknown as well as their potential to better predict the outcome of patients. To address this question, time must be considered as a formal variable of the problem and not just a simple repeated observation. Here, we hypothesize that the temporal formation of basic connection blocks such as intermodule edges and intramodule triangles would be sufficient to determine the large-scale brain reorganization after stroke. To test our hypothesis, we adopted a statistical approach based on temporal exponential random graph models (tERGMs). First, we validated the overall performance on synthetic time-varying networks simulating the reconfiguration process after stroke. Then, using longitudinal functional connectivity measurements of resting-state brain activity, we showed that both the formation of triangles within the affected hemisphere and interhemispheric links are sufficient to reproduce the longitudinal brain network changes from 2 weeks to 1 year after the stroke. Finally, we showed that these temporal connection mechanisms are over-expressed in the subacute phase as compared to healthy controls and predicted the chronic language and visual outcome respectively in patients with subcortical and cortical lesions, whereas static approaches failed to do so. Our results indicate the importance of considering time-varying connection properties when modeling dynamic complex systems and provide fresh insights into the network mechanisms of stroke recovery.

Motivation & Objective

• To investigate whether dynamic, time-varying network mechanisms underlie functional reorganization after stroke.
• To determine if temporal connection patterns—such as triangle formation and interhemispheric links—can predict chronic functional outcomes.
• To compare the predictive power of temporal network models against static network approaches in stroke recovery.
• To validate the proposed model on synthetic networks simulating post-stroke reconfiguration.

Proposed method

• Applied temporal exponential random graph models (tERGMs) to analyze longitudinal functional connectivity (FC) networks derived from resting-state fMRI data collected at 2 weeks, 3 months, and 1 year post-stroke.
• Constructed time-varying, sparse adjacency matrices by thresholding FC matrices at 10% connection density to ensure comparability across subjects.
• Restricted model specifications to focus on triangle formation within the affected hemisphere and interhemispheric link formation, using a product term to enforce longitudinal closure of triangles.
• Validated model performance on synthetic time-varying networks mimicking post-stroke network reconfiguration.
• Compared tERGM results in stroke patients to demographically matched healthy controls to identify over-expressed temporal mechanisms.
• Used a modular network approach, modeling only the brain regions affected by the lesion to improve specificity, especially for cortical and subcortical lesions.

Experimental results

Research questions

• RQ1 Do temporal connection mechanisms such as triangle formation and interhemispheric link development predict post-stroke functional recovery better than static network measures?
• RQ2 Are the temporal dynamics of brain network reorganization in the subacute phase over-expressed in stroke patients compared to healthy controls?
• RQ3 Can temporal network signatures derived from early post-stroke networks predict chronic functional outcomes in patients with cortical or subcortical lesions?
• RQ4 Do time-varying network properties capture higher-order network reorganization patterns that static models miss?
• RQ5 Is the formation of intrahemispheric triangles and interhemispheric connections sufficient to reproduce longitudinal changes in brain network topology after stroke?

Key findings

• The formation of intrahemispheric triangles within the affected hemisphere and interhemispheric links were sufficient to reproduce longitudinal brain network changes from 2 weeks to 1 year post-stroke.
• These temporal connection mechanisms were significantly over-expressed in the subacute phase (2 weeks to 3 months) compared to healthy controls.
• tERGM-based temporal signatures predicted chronic language and visual outcomes in patients with subcortical and cortical lesions, respectively, whereas static network approaches failed to do so.
• The model successfully captured network reorganization dynamics using a sparse, thresholded adjacency matrix at 10% connection density, ensuring robust cross-subject comparison.
• The inclusion of a product term in the tERGM model ensured that triangle closure occurred only over time, enforcing longitudinal formation rather than instantaneous appearance.
• For subcortical lesions, the affected hemisphere was defined based on lesion location—lesion side for white matter, contralesional for cerebellar, and both for brainstem lesions—improving model specificity.