Ricerc@Sapienza

Hybrid Brain-Computer Interfaces (BCIs) for upper limb rehabilitation after stroke should enable the reinforcement of 'more normal' brain and muscular activity. It is well known that the regaining of motor function after stroke is characterized by several changes in muscular activation patterns, such as motor overflow, co-activation of agonist and antagonist muscles and spasticity. Hybrid BCIs include peripheral signals such as those derived from electromyography (EMG) as a control feature. These have mostly been developed to improve the classification performance of the system e.g. in assistive BCIs, with little or no focus on which properties of the EMG signals should be considered in a rehabilitative context.
This proposal aims at developing a hybrid BCI prototype in which the control features will be derived from a combined Electroencephalographic (EEG) and EMG connectivity pattern estimated online during upper limb attempts and used to drive a feedback to the user through Functional Electrical Stimulation (FES) when 'correct' movements are detected.
To achieve this goal, pre-recorded EEG and EMG data of 20 healthy subjects and 20 stroke patients during simple upper limb movements/attempts will be analysed to characterize physiological and pathological patterns. The best individual control features able to detect the physiological efferent drive and the dysfunctional patterns will be selected, a hybrid classification approach will be built and the ability of these hybrid features to classify the movements in real-time will be tested. A novel BCI prototype will be developed and used to control a FES device. Upon success, this prototype could pave the way towards a novel hybrid BCI system for post-stroke rehabilitation based on a global vision of the physiological/pathological patterns involved in the movement that has to be recovered.