Stroke is known to disrupt large scale functional networks in the brain. More detailed characterizations of this network dysfunction will have important implications for mechanistically informed interventions to enhance recovery. This study tested whether lesion distribution explains variation in post-stroke functional connectivity (FC) and evaluated a parametric mean field model (pMFM; Kong et al., 2021) as a tool for simulating lesion effects on cortical dynamics in stroke. FC was computed from resting-state fMRI using the CONN Toolbox for 30 chronic stroke patients with aphasia. Manually drawn lesion maps were overlaid on the Desikan–Killiany (DK) anatomical parcellation and JHU ICBM-DTI-81 white matter tract atlas. DK regions with >50% overlap were considered lesioned. Tract disruptions were quantified based on lesion-tract overlap and used to weight disconnections between region pairs. Individualized structural connectivity (SC) matrices were generated by modifying the group-averaged SC derived by Kong et al. (2021) from Human Connectome Project diffusion data. Simulations of neural dynamics were run using the pMFM with lesion-modified SC and published model parameters. Lesion simulations showed significant decreases in FC, particularly in regions most closely connected to lesioned regions determinded by minimum shortest path length to a lesioned region (Wilcoxon signed rant test, p=1.9e-9). Experimental FC showed a similar pattern with significantly reduced FC in regions closely connected to the lesion (Wilcoxon signed rant test, p=1.9e-9), but no significant difference in distant regions (Wilcoxon signed rant test, p=0.94). Additionally, simulated edgewise FC significantly predicted experimental FC (linear mixed effects model, p<2e-16, random effects of subject and edge). The lesion-based neural mass model used in this study accurately captured patterns of network dysfunction in individuals with post-stroke aphasia. Further development of this model will facilitate testing of hypotheses about mechanisms of network dysfunction as well as recovery after stroke, with the ultimate goal of developing new interventions, including mechanistically informed neurostimulation techniques to enhance recovery. References: Kong, R., Li, J., Orban, C. et al. (2021). Spatial topography of individual-specific cortical networks predicts human cognition, personality, and emotion. Nature Communications , 12, 6042.
Falconer et al. (Thu,) studied this question.