Effective connectivity extracts clinically relevant prognostic information from resting state activity in stroke

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Imaging

Background:

• Resting-state fMRI reveals biomarkers for acute brain dysfunction post-stroke, including reduced inter-hemispheric and increased ipsi-lesional functional connectivity.
• Previous whole-brain modeling showed reduced integration and segregation in stroke patients compared to healthy individuals.

Purpose of the Study:

• To develop and validate a novel method for inferring whole-brain directional effective connectivity from resting-state fMRI data.
• To compare the predictive accuracy of effective connectivity versus functional connectivity for stroke status, patient performance, and long-term outcomes.

Main Methods:

• Inferred whole-brain directional effective connectivity from zero-lagged and lagged covariance matrices.
• Compared functional connectivity and model-based effective connectivity in predicting stroke vs. healthy status and patient deficits.
• Assessed prediction accuracy at multiple time points post-stroke (1-2 weeks, 3 months, 1 year).

Main Results:

• Both functional and effective connectivity predicted stroke status better than chance.
• Effective connectivity showed significantly higher accuracy than functional connectivity in predicting stroke status and long-term outcomes at all measured time points.
• Effective connectivity better predicted the number of patient deficits, with early effective connectivity values predicting later deficits.

Conclusions:

• Whole-brain directional effective connectivity derived from resting-state fMRI provides crucial information for clinical prognosis in stroke patients.
• This novel method offers superior predictive power for stroke outcomes compared to traditional functional connectivity measures.
• Early effective connectivity patterns are valuable for predicting long-term neurological deficits after stroke.