Interactive rehabilitation and dynamical analysis of scalp EEG.

Electroencephalography (EEG) has been used for decades to measure the brain's electrical activity. Planning and performing a complex movement (e.g., reaching and grasping) requires the coordination of muscles by electrical activity that can be recorded with scalp EEG from relevant regions of the cortex. Prior studies, utilizing motion capture and kinematic measures, have shown that an augmented reality feedback system for rehabilitation of stroke patients can help patients develop new motor plans and perform reaching tasks more accurately. Historically, traditional signal analysis techniques have been utilized to quantify changes in EEG when subjects perform common, simple movements. These techniques have included measures of event-related potentials in the time and frequency domains (e.g., energy and coherence measures). In this study, a more advanced, nonlinear, analysis technique, mutual information (MI), is applied to the EEG to capture the dynamics of functional connections between brain sites. In particular, the cortical activity that results from the planning and execution of novel reach trajectories by normal subjects in an augmented reality system was quantified by using statistically significant MI interactions between brain sites over time. The results show that, during the preparation for as well as the execution of a reach, the functional connectivity of the brain changes in a consistent manner over time, in terms of both the number and strength of cortical connections. A similar analysis of EEG from stroke patients may provide new insights into the functional deficiencies developed in the brain after stroke, and contribute to evaluation, and possibly the design, of novel therapeutic schemes within the framework of rehabilitation and BMI (brain machine interface).