Functional topology in the acute stroke phase

After constructing the functional connectivity networks using partial correlation, we explore the characteristic path length for positive, negative and absolute weights separately. Fig. \ref{405899} illustrates a positive correlation between the early follow-up NIHSS scores and the characteristic path length after preserving all positive connections of the functional network with nodes defined by the Harvard-Oxford atlas (Pearson's correlation coefficient \(r=0.42, p=0.01\)). A similar relationship can be observed with the AAL atlas, for which  Pearson's correlation coefficient is \(r=0.38, p=0.03\), as well as with the Destrieux atlas, with Pearson's correlation coefficient \(r=0.41,p=0.02\). This finding could potentially indicate that the shorter the characteristic path length in the acute phase of stroke, the more resilient the functional network is to brain damage; hence, the better the functional outcome as measured using NIHSS.
We further assessed the differences in functional network measures with respect to the 'good' or 'poor' mRS outcome. Fig. \ref{366792} demonstrates the differences identified between \(L\) for patients with 'good' (69% of patients) and 'poor' (31% of patients) outcome at 90 days post-stroke. This result is consistent across all three atlases. Our finding indicates that the characteristic path length estimated from rsfMRI scans in the acute stroke phase can be used as a predictor of functional recovery at a later stage.