Using in vivo functional and structural connectivity to predict chronic stroke aphasia deficits

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Neurology

Background:

• Focal brain damage from stroke can cause aphasia.
• Aphasia may stem from network-level brain disorders, not just localized damage.
• Previous research explored brain-behavior relationships using lesion data, but few integrated in vivo structural and functional connectivity.

Purpose of the Study:

• To determine if multimodal neuroimaging (structural and functional connectivity) improves prediction of language-cognitive factors in post-stroke aphasia compared to lesion models alone.
• To assess the predictive power of neural measurements for phonology, semantics, executive function, and fluency.

Main Methods:

• Structural and functional magnetic resonance imaging (MRI) were used in patients with chronic post-stroke aphasia (n=68 for structural, n=39 for functional).
• Models were constructed to predict four language-cognitive factors.
• Regularized regression models were employed, with varying sparsity hyperparameters.

Main Results:

• Each neural measurement individually related significantly to phonology, semantics, and fluency, but not executive function.
• Structural and functional connectivity models did not explain additional variance beyond lesion models.
• Predictive functional connectivity features were within resting-state networks; predictive structural connectivity features were within lesion sites.

Conclusions:

• Network-level disruption in post-stroke aphasia is predictable by lesion location alone.
• Integrating structural and functional connectivity did not enhance prediction models for language impairment.
• Future studies should optimize sparsity hyperparameters for different behavioral components and imaging features.