Multivariate Connectome-Based Symptom Mapping in Post-Stroke Patients: Networks Supporting Language and Speech

Area of Science:

• Neuroscience
• Neurolinguistics
• Computational Psychiatry

Background:

• Language processing involves a complex network of brain regions.
• Traditional stroke lesion mapping may miss crucial information about network disconnections.
• Existing methods struggle to differentiate between brain regions with shared vascular supply.

Purpose of the Study:

• To investigate the relationship between post-stroke aphasia and structural brain damage using a multivariate predictive framework.
• To compare the accuracy of lesion maps versus structural connectome integrity in predicting language deficits.
• To identify specific brain networks and connections critical for language function and recovery.

Main Methods:

• Applied a multivariate predictive framework to analyze lesion maps and structural connectome data in 90 individuals with chronic post-stroke aphasia.
• Utilized a cross-validation framework to predict language scores from the Western Aphasia Battery (WAB).
• Compared predictive accuracy of lesion-based and connectome-based analyses.

Main Results:

• Both lesion mapping and connectome analysis achieved comparable, better-than-chance prediction accuracy for all WAB scores.
• Connectome-based analysis revealed distinct, overlapping cortical networks implicated in specific speech functions.
• Highlighted the critical role of temporoparietal junction connectivity for language tasks, independent of lesion location.

Conclusions:

• Multivariate connectome analysis is a valuable complement to traditional lesion mapping for understanding aphasia.
• Brain connectivity analysis offers insights into language networks, even in regions spared from direct tissue damage.
• Combining lesion and connectome approaches provides a more comprehensive understanding of aphasic impairment and recovery.