Brain-Machine Interface Might Aid Poststroke Rehabilitation

S. Andrew Josephson, M.D.


AccessMedicine from McGraw-Hill 

Patients following stroke may be left with residual weakness in the limbs despite intensive physical therapy. A number of strategies to improve function in these patients have been described recently, including constraint-induced therapy. However, patients with severe weakness will often have difficulties with many of the techniques that require some degree of muscle strength in order to be successful. Ramos-Murguialday and colleagues (2013) aimed to test a novel brain-machine interface paradigm in the hopes of improving strength in patients with chronic severe weakness following stroke.

The authors enrolled 32 patients who had severe weakness of at least 1 hand with no active finger extension of at least 10 months' duration. The intervention tested was brain-machine interface training in which the patient would learn to control desynchronization of sensorimotor rolandic brain oscillations as measured by electroencephalography. In the intervention group, a patient's successful control of these oscillations was translated into immediate movement of an orthosis that allowed for limb grasping and reaching movements. In the control or "sham" group, the same brain-machine interface training occurred, but oscillations were randomly linked with movements of the orthosis, regardless of success. Physical therapy followed each brain-machine interface training session for both groups, and the groups participated in a mean of just under 18 days of training and therapy.

The authors found a significant group-versus-time interaction in standard assessments of upper limb motor scores; these scores improved significantly more in the intervention group than in the sham group, demonstrating a clinically relevant improvement from zero to some activity in paretic musculature. Only patients in the intervention group were able to significantly improve their brain-machine interface performance over time. Analysis of electromyography (EMG) in the affected limb showed a statistically significant improvement in the intervention group before and after training that was not seen in the sham group. Additional investigations looked at functional MRI (fMRI) measures of blood oxygenation level–dependent activity before and after intervention; there was found to be improvement in laterality index corresponding to the improvements seen in motor scores in the treatment group. Comparable scores were found in the two groups on placebo-expectancy measures, suggesting in part that the participants were reasonably blind to their assigned treatment group.

This study demonstrates that a brain-machine interface can be used to improve motor scores in poststroke patients even well after their initial injury (>10 months). The model here is one of training to improve proprioceptive positive feedback that can then in theory "prime" the brain and limbs for the beneficial effects of physical therapy training sessions. The sham portion of the study importantly demonstrates that the physical therapy training sessions themselves were not the source of the observed improvement. This is certainly only a proof-of-concept study and requires additional confirmation in order to determine if indeed this (or some other similar) approach can provide a desperately needed boost to poststroke rehabilitation outcomes.