When using virtual reality (VR), observation of an action, even simulated, allows recruitment of stored motor programs. These processes are expressed by changes in α and β oscillation magnitude on electroencephalography across brain areas belonging to the mirror neuron system. This study investigated the neurophysiological underpinning of gait recovery induced by the observation of an animated avatar while performing robot-assisted gait training.

This pilot study included patients at least 55 years of age with spasticity secondary to a first ever ischemic stroke of at least six months′ duration. All subjects received Locomat training, either with or without an animated avatar, in a two-dimensional VR. The participants were assessed for clinical, kinematic and electroencephalogram changes. The primary outcome variable was the proportion of patients achieving 20% improvement in lower limb gait and balance, as measured with the Rivermead Mobility Index (RMI), the Tinetti Performance of Oriented Mobility Assessment (POMA) and the gait cycle-related Event-Related-Spectral- Perturbations (ERSPs).

The RMI improved significantly more in the VR group than in the control group (P<0.001), while the POMA improved only in the VR group (P＝0.001). In addition, hip force and knee force improved significantly more in the VR group (P＝0.02 for both comparisons). Significant relationships were found between RMI and POMA score improvements and EEG measures of central ERSPs (P＝0.001 and P＝0.04, respectively) and frontal Hγ-ERD magnitude (P＝0.003). Improvement in hip force was significantly related to frontal Hγ-ERD magnitude (P＝0.004).

This study found that virtual reality during gait training can elicit stronger correlations in the fronto-parietaloccipital areas involved in motor intention and planning.