Network Connectivity Measures to Assess Changes in Brain Activity of Stroke Patients

Most stroke survivors present hemiplegia or hemiparesis, which impair severely upper limb motor function and independence. Clinical scales are commonly used to evaluate the effects of post-stroke rehabilitation programs. However, those scales rely on subjective assessment of patient's behavior or physical state, requiring highly trained specialists. In this work, the use of basic connectivity parameters based on graph theory applied to EEG signals, was explored as a quantitative tool to evaluate the effects in brain activity of an upper limb rehabilitation intervention. The approach was tested on a group of six stroke patients, using routine, resting-state EEG signals recorded from frontal, central and parietal regions, before and after their participation in an upper limb rehabilitation intervention using a P300-based Brain Computer Interface coupled to a Functional Electrical Stimulator. Brain connectivity was estimated through two complex parameters: node degree and clustering coefficient, calculated from two phase synchrony metrics (weighted phase lag index and spectral coherence) in the alpha (8–12 Hz) and lower beta (13–19 Hz) frequency bands. After the intervention, the average node degree for these patients was increased at central and frontal EEG electrodes, and the average clustering coefficient was increased at parietal and central positions. Furthermore, clinical scales showed improvements in upper limb function after the intervention. The next steps are to analyze the relationships between clinical and brain connectivity measures, and to follow up the patient brain and clinical state at mid and long-term spans.