Detecting brain network changes induced by a neurofeedback-based training for memory function rehabilitation after stroke

The efficacy of rehabilitative interventions in stroke patients is routinely assessed by means of a neuropsychological test battery. Nowadays, more evidences indicate that the neuroplasticity which occurs after stroke can be better understood by investigating changes in brain networks. In this pilot study we applied advanced methodologies for effective connectivity estimation in combination with graph theory approach, to define EEG derived descriptors of brain networks underlying memory tasks. In particular, we proposed such descriptors to identify substrates of efficacy of a Brain-Computer Interface (BCI) controlled neurofeedback-based intervention to improve cognitive function after stroke. EEG data were collected from two stroke patients before and after a neurofeedback-based training for working memory deficits. We show that the estimated brain connectivity indices were sensitive to different training intervention outcomes, thus suggesting an effective support to the neuropsychological assessment in the evaluation of the changes induced by the BCI-based rehabilitative intervention.