ABSTRACTRecent resting-state fMRI studies in stroke patients have identified two robust biomarkers of acute brain dysfunction: a reduction of inter-hemispheric functional connectivity (FC) between homotopic regions of the same network, and an abnormal increase of ipsilesional FC between task-negative and task-positive resting-state networks (RSNs). Whole-brain computational modeling studies, at the individual subject level, using undirected effective connectivity (EC) derived from empirically measured FC, have shown a reduction of measures of integration and segregation in stroke as compared to healthy brains. Here we employ a novel method, first, to infer whole-brain directional EC from zero-lagged and lagged FC, then, to compare it to empirically measured FC for predicting stroke vs. healthy status, and patient performance (zero, one, multiple deficits) across neuropsychological tests. We also investigated the accuracy of FC vs. model EC in predicting the long-term outcome from acute measures.Both FC and EC predicted healthy from stroke individuals significantly better than the chance-level, however, EC accuracy was significantly higher than that of FC at 1-2 weeks, three months, and one year post-stroke. The predictive FC links mainly included those reported in previous studies (within-network inter-hemispheric, and between task-positive and -negative networks intra-hemispherically). Predictive EC links included additional between-network links. EC was a better predictor than FC of the number of behavioral domains in which patients suffered deficits, both at two weeks and one-year post onset of stroke. Interestingly, patient deficits at one-year time point were better predicted by EC values at two weeks rather than at one-year time point. Our results thus demonstrate that the second-order statistics of fMRI resting-state activity at an early stage of stroke, derived from a whole-brain EC, estimated in a model fitted to reproduce the propagation of BOLD activity, has pertinent information for clinical prognosis.
Decreased integration and information capacity in stroke measured by whole brain models of resting state activity