Abstract:
Introduction: Brain–machine interfaces (BMI) create a direct communication pathway between the brain and an external device. BMI can be used, beside other possibilities, for selective induction of use-dependent neuroplasticity that might e.g. facilitate motor recovery. Objectives: The dissertation pursued four main goals: 1) to investigate the efficacy of BMI technology as a rehabilitation tool for chronic stroke patients suffering complete paralysis of their fingers and from a damaged brain at the same time; 2) to find biomarkers, e.g. the presence of motor evoked potentials (MEP) that can predict recovery related to BMI training; 3) to investigate the neural substrates, e.g. integrity of the cerebral cortex or thalamus to generate event-related desynchronization (ERD) and thus to control the BMI, and 4) to integrate other biosignals to improve BMI control. Methods: 39 severely affected chronic stroke patients with no finger extension underwent a 4-week daily BMI training for one and half hour followed by one hour of physiotherapy. Patients were divided according to feedback contingency, and subcategorized according to the integrity of sensorimotor cortex, thalamus and presence of MEP. Results: The results show that patients in the experimental group improved functional outcomes significantly compared to the control group. Patients with ipsilesional upper-limb MEP presented better functional outcomes in both treatment groups, but motor recovery was superior in patients with MEP in the experimental group. Besides that, patients with an intact motor cortex showed significantly stronger ERD since their first training day. Moreover integration of electrooculogram (EOG) seems to improve reliability of BMI control. Interpretation: The results show that BMI technology is a reliable tool in neurorehabilitation of chronic stroke patients. The success of BMI training can be improved according to the integrity of the motor cortex or the presence of MEP.