Online EEG Analysis for Neurofeedback in Post-Stroke Neurorehabilitation of the Lower Limbs

• Applicant: Professor Dr.-Ing. Jens Haueisen
• Subject Area: Clinical Neurology; Neurosurgery and Neuroradiology ; Biomedical Systems Technology ; Cognitive, Systems and Behavioural Neurobiology ; Medical Physics, Biomedical Technology
• Term: since 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 397686322

Stroke is one of the leading causes of disability worldwide with more than 24 million new cases per year. While the age-standardized mortality rates have decreased significantly over the last decades, the decrease in incidence rate has been less pronounced. Therefore, the improvement of post-stroke rehabilitation remains of high priority. Particularly for post-stroke motor impairment it was shown that rehabilitation training is highly effective. Electroencephalography (EEG) is unique amongst neuroimaging methods, as it is the only non-invasive methods that offers a time resolution in the millisecond range while being highly mobile, easy to apply, and comparably cheap. These features make EEG an attractive tool for brain-computer interfaces (BCI). BCIs are promising tools in rehabilitation as they provide direct visual and sensory feedback about the success of a mental task. Within this research project, we will explore and advance the use of BCIs based on EEG in combination with a rehabilitation robot in post-stroke lower limb motor rehabilitation. To date, a focus of the application of BCIs in post-stroke rehabilitation has been on upper limb motor rehabilitation. The application in lower limb motor rehabilitation has been less explored due to the more complex signal processing. Based on novel algorithms to detect motor imagery (MI) we aim to advance the use of BCIs in lower limb motor rehabilitation. We will research novel classification algorithms based on convolutional neural networks incorporating online EEG source and connectivity analysis to improve the classification accuracy of lower-limb MI. We will acquire EEG data of healthy subjects and patients during passive movement and MI of the lower limbs. For the passive movement, we will apply a novel rehabilitation robot based on an end-effector robotic arm with six degrees of freedom, which enables complex trajectories and exactly timed repetitions. We will evaluate the obtained EEG data and exploit the results in the development of the novel classification algorithms. We will perform a proof-of-concept experiment to demonstrate the use of the rehabilitation robot to provide direct sensory feedback in a closed-loop scenario. The novel classification algorithms will improve the classification accuracy in all applications of MI, especially for BCIs. The evaluation of the measurement data will increase the knowledge about the sources of brain activity and the connectivity during complex passive movement and MI of the lower limbs. The application in a closed-loop scenario with BCI and a rehabilitation robot has the potential to significantly improve post-stroke rehabilitation of lower limb motor impairments.

• DFG Programme: Research Grants
• International Connection: Austria
• Partner Organisation: Fonds zur Förderung der wissenschaftlichen Forschung (FWF)
• Cooperation Partner: Professor Dr.-Ing. Daniel Baumgarten