Post-stroke outcomes predicted from multivariate lesion-behaviour and lesion network mapping.

Area of Science:

• Neuroscience
• Neurology
• Medical Imaging

Background:

• Accurate prediction of chronic post-stroke cognitive and motor outcomes remains difficult due to high inter-individual variability and reliance on clinical heuristics.
• Neuroanatomical location of a stroke is a key predictor of long-term outcomes, offering an opportunity for empirically based predictions using clinical neuroimaging data.

Purpose of the Study:

• To evaluate the predictive power of lesion-behavior maps for 12-month post-stroke outcomes in an independent sample.
• To assess if augmenting lesion-behavior maps with structural and functional lesion network mapping improves outcome prediction.

Main Methods:

• Leveraged lesion-behavior maps from two large cohorts to predict 12-month outcomes in an independent stroke patient sample.
• Utilized structural and functional lesion network mapping, seeding from peak lesion-deficit regions, to elucidate disrupted networks.
• Quantified prediction variance using latent growth curve models, comparing lesion-behavior mapping with lesion network mapping.

Main Results:

• All lesion-deficit mapping modalities significantly predicted variance in 12-month cognitive and motor outcomes.
• Structural and functional lesion network mapping provided predictive power beyond traditional lesion-behavior mapping.
• Functional lesion network mapping excelled in predicting language deficits, while structural mapping was superior for motor deficits.

Conclusions:

• Combining lesion location with lesion network mapping enhances the prediction of post-stroke deficits at 12 months.
• Lesion network mapping offers a more precise approach to understanding and predicting stroke recovery compared to single-lesion analyses.