ECoG Processing for Motion Kinematics Reconstruction

Nowadays various neural interfaces are currently being developed in order to facilitate interaction between a human and a computer. Bidirectional neural interfaces are especially interesting because they allow transferring both information from person to computer and in the opposite direction. In particular, such an interface could allow a person facing problems with movement to process signals sent by the brain and send back with a predicted action (which can be transmitted, for example, with a robotic arm). In this work, using electrocorticography data of the patient and data with movements of his fingers, we solved the problem of predicting movements of a person’s fingers according to the activity of his brain. The results of the application of machine learning algorithms showed that with an accuracy of more than 75% and a relatively low error, one can predict human movement based on the ECoG results.