Parallel ICA reveals linked patterns of structural damage and fMRI language task activation in chronic post-stroke aphasia

TL;DR

This study uses parallel ICA to link structural brain damage and functional activation patterns in post-stroke aphasia, revealing how damage impacts language network recruitment and compensation mechanisms.

Contribution

It introduces a novel application of parallel ICA to connect structural damage with functional activation patterns in stroke patients, highlighting specific lesion-behavior relationships.

Findings

Damage to left posterior temporo-parietal cortex disrupts language pathways.

Right hemispheric activation increases compensate for left-sided damage.

Lesion severity correlates with poorer language performance.

Abstract

Structural and functional MRI studies of patients with post-stroke language deficits have contributed substantially to our understanding of how cognitive-behavioral impairments relate to the location of structural damage and to the activation of surviving brain regions during language processing, respectively. However, very little is known about how inter-patient variability in language task activation relates to variability in the structures affected by stroke. Here, we used parallel independent component analysis (pICA) to characterize links between patterns of structural damage and patterns of functional MRI activation during semantic decisions. The pICA analysis revealed a significant association between a lesion component featuring damage to left posterior temporo-parietal cortex and the underlying deep white matter and an fMRI component featuring (1) heightened activation in a primarily right hemispheric network of frontal, temporal, and parietal regions, and (2) reduced activation in areas associated with the semantic network activated by healthy controls. Stronger loading parameters on both the lesion and fMRI activation components were associated with poorer language test performance. Fiber tracking suggests that lesions affecting the left posterior temporo-parietal cortex and deep white matter may lead to the simultaneous disruption of multiple long-range structural pathways connecting distal language areas. Damage to the left posterior temporo-parietal cortex and underlying white matter may (1) impede the language task-driven recruitment of canonical left hemispheric language and other areas (e.g. the right anterior temporal lobe and default mode regions) that likely support residual language function after stroke, and (2) lead to the compensatory recruitment of right hemispheric fronto-temporo-parietal networks for tasks requiring semantic processing.