The present thesis investigates age-related and stroke-induced differences in brain activation and neural network configuration, during states of rest and states of attentional demand. To this end, we employed a multiple object tracking (MOT) task at two load levels and compared functional magnetic resonance imaging between a group of younger and a group of older healthy participants (papers I and II), and between a group of sub-acute stroke patients and healthy controls (paper III).
In paper I, we probe the age-related differential network response during rest and attentional demand. Task-positive (TPN) and task-negative networks (TNN) represent functionally connected brain regions that reliably activate and deactivate, respectively, when performing attention-demanding tasks. The antagonistic relationship between these networks has been hypothesized to reflect ongoing regulation of cognitive control and task-related effort, and as such a prerequisite for efficient visuospatial attention. In the field of cognitive aging, attentional impairment is widely reported, but it has not been known to which degree these behavioral differences can be attributed to network-specific neuronal alterations in the TPN and task-negative networks TNN, respectively. Therefore, by employing independent component analysis (ICA) to derive brain networks from fMRI data recorded during a blocked MOT task, which requires sustained multifocal attention, we sought to characterize age-related alterations in TPN and TNN during sustained visuospatial attention. The results demonstrated age-related differences in network response during MOT. We found for the old compared to the young group diminished activations and deactivations in the task-positive and task-negative network, respectively. Moreover, increasing task difficulty resulted in a diminished network response in the old compared to the young group. Performance level, as indicated by target detection accuracy, was higher in the young compared to the old group. Assessment of network co[1]activation showed for the younger group stronger correlations within networks that were designated as TPN and TNN, compared to the older group. Summarily, the findings supported the notion of neural dedifferentiation with increasing age.
In paper II, we again studied age-related differences during MOT using roughly the same study sample as in paper I (based on task performance, a small subset of participants in paper I were excluded in paper II and vice versa). Whereas in paper I we investigated task-related network activation and co-activation, in paper II we probed age-related differences in functional connectivity (FC) across rest and task. We utilized a machine learning classifier to investigate whether age-related connectivity changes become more apparent during cognitive engagement. Variability in brain activity reflects optimal brain function. To investigate age related differences in signal variability, we computed the standard deviation of signal amplitude (SDSA) across network nodes and performed groupwise comparisons within each cognitive condition. Results from the study revealed robust discrimination between cognitive states across age groups. Further, there was a load dependent increase in classification accuracy between age groups when the participants were task-engaged compared to the resting state condition. Thus, confirming that the MOT task-paradigm increases sensitivity to age-related differences in functional connectivity compared to an unconstrained resting state. Network nodes within the dorsal attention network (DAN) and the default mode network (DMN) – representing the main task-positive and task[1]negative networks, respectively – were shown to be the nodes most sensitive to age effects. These networks have an anti-correlated relationship which were found reduced in the older group comparative to the younger group, corroborating the neural dedifferentiation theory and converging with the findings in paper I. Overall, signal variability was higher for the younger group, and nodes within the DAN and DMN showed strongest the effects of task.
In paper III, we used the same methodological framework as in paper II, this time to investigate functional connectivity differences between a group of healthy controls and a group of sub-acute ischemic stroke patients. We again showed high classification accuracy between resting state and the two load levels of the MOT task. However, machine learning classification did not discriminate stroke patients from healthy controls beyond chance-level accuracy. The inherent demands of the study selected for a patient group with clinically mild strokes, and fMRI findings together with behavioral data converged to suggest that mild strokes might impart sparse effects on cognitive function and neural networks beyond the lesion proper