Introduction: Despite its prevalence in stroke research, the vascular dependence of functional MRI (fMRI) makes it susceptible to changes in factors such as vascular reactivity that may occur over the course of stroke recovery, potentially confounding longitudinal findings. Scaling fMRI data with respect to vascular properties may account for some of the high variability often observed in patterns of motor task activation throughout recovery.
Hypothesis: We hypothesized that scaling fMRI data by the amplitude of resting-state fluctuations (RSFA), an estimate of vascular reactivity, would reduce intersubject variability of motor task activation in both patients and similarly-aged controls while minimizing between-group differences and revealing a more consistent pattern of changes in activation associated with stroke recovery.
Methods: 24 ischemic stroke patients (ages 44-84 years, mean = 64) were scanned at early (2-12 days, mean = 5) and late (mean = 11 months) stages after onset. 20 healthy controls (ages 47-74 years, mean = 59) were also scanned, 11 of whom were scanned again after ~6 months. Subjects performed two 20-second-block finger-tapping tasks, one with each hand, and a 10-minute resting-state fMRI scan, from which the scaling factor (RSFA) was computed. Percent signal change and activation volume before and after scaling were measured across several bilateral motor areas and compared between groups and within each group longitudinally.
Results: Scaling with RSFA significantly reduced intersubject variability of intensity and extent of activation in both groups while mitigating between-group differences in activation in multiple motor areas. Further, scaling revealed a significant reduction of activation across motor regions in stroke patients over time that was not apparent without scaling. This pattern more closely resembled that of controls, for which it was evident both before and after scaling.
Conclusions: Our findings point to a role for fMRI calibration in longitudinal studies of plasticity and recovery after stroke, and suggest that the large variability often observed in activation patterns associated with recovery may in part be vascular in origin.