Reshaped functional connectivity gradients in acute ischemic stroke

Abstract

Ischemic brain stroke disrupts blood flow, leading to functional and structural changes associated with behavioral deficits. Importantly, despite this disruption occurring in localized regions, the resulting changes in the functional organization are both high-dimensional and widespread across the human cortex. However, the mechanisms with which these global patterns emerge and the subsequent behavioral deficits they entail, remain largely unexplored. Functional connectivity gradients provide consistent, reproducible, and robust low-dimensional representations of brain function that can be explored to reduce brain heterogeneity to a handful of axes along which brain function is organized. Here, we investigated how stroke disrupts this canonical gradient space by aligning each patient to a control-averaged gradient embedding and computing the distances to the "correct" positions to quantify functional deviations and their contribution to behavioral deficits. Importantly, we explicitly corrected these gradients for stroke-induced hemodynamic lags to further study their contribution. We found that lag correction enhanced the functional connectivity gradients most prominently in the second gradient, on which visual and somatomotor function is concentrated. Additionally, we identified significant functional deviations primarily within somatomotor, visual, and ventral attention networks, correlating with behavioral impairments. We studied the hemispheric asymmetries of these deviations finding that intact hemispheres preserve comparable patterns of asymmetry while damaged ones presented important changes. Lastly, right-sided lesions displayed more localized functional deviations than their contralateral lesions. Overall, we provide evidence that 1) correcting for hemodynamic lags improves gradient accuracy, as indicated by increased percentages of explained variance, and 2) behavioral impairments and hemispheric asymmetries result from a repositioning of region-based connectivity profiles in a low-dimensional, interpretable space. This suggests that large-scale brain function alterations manifest in slight, predictable movements largely confined to the visual-somatomotor axis.