Since the beginning, the focus of the work has been to implement and test the framework of the novel brain-computer interface (BCI) platform. This part of the project was performed while at the Institute of Neurorehabilitation Systems (NRS, Göttingen, Germany). The platform consisted of a desktop computer, the EEG recording system and of a sensorised pedal device for facilitating the recovery of ankle function of an ankle foot orthosis. These devices were controlled via software and allowed the recording of EEG signals during actual and imagined movements. Movement-related cortical potentials (MRCPs) were then extracted (MRCPs). They consist in a low-frequency negative shift (0.05- 3 Hz) in the EEG signals occurring 2 s ahead a voluntary movement. The MRCP thus, reflect the cortical processes involved in movement planning and movement preparation. The novel added value to the platform resides in the addition of a system for recording high-density EMG signals (HD-EMG). All the elements in the platform were synchronized and the final system was tested on a cohort of 30 acute stroke patients recruited at the Neurorehabilitation unit of the Brøndelsev Hospital during the secondment period at Aalborg University (AAU, Aalborg, Demark). Patients were divided into two groups (one BCI and one SHAM), in a blind way to the patient, the physicians and the therapists and they underwent a 12 sessions training, during which the EEG was recorded while patients performed an ankle dorsiflexion using the sensorised pedal. The intention of movement was detected from EEG signals and electrical stimulation was delivered at motor threshold to the tibial nerve at the time of occurrence of the negative peak of the MRCP. Patients within the SHAM group underwent the same training, but in this case the stimulation they received was at sensory level. The platform was further improved, by adding the system for Transcranial Magnetic Stimulation to be used before and after the training of the very first and the very last sessions to assess the changes in the motor potentials evoked by the magnetic field. On the first and twelfth session, HD-EMG was also recorded, before and after the BCI training to assess the changes in corticospinal pathways and motor neuron response to training. Some features were extracted from the HD-EMG and the coherence between motor neuron discharge timings was computed. In fact, coherence function provides information about the oscillations generated by the nervous system and is commonly used to quantify the strength and the frequency of common synaptic inputs to the motor neuron pool as well as to trace the neuronal reorganizations functionally relevant for motor recovery. The results show that at baseline, acute stroke survivors exhibit a coherence peak around 13 Hz, which cannot be identified in the coherence spectrum computed in a cohort of healthy age-matched individuals. This result outlines how the motor system reorganizes shortly after a stroke and how the therapies for the recovery of movement should be tailored to allow for a more functional and rapid recovery.
