Patient-Specific Robot-Assisted Stroke Rehabilitation Guided by EEG - A Feasibility Study

Authors

Maryam Butt, University of WollongongFollow
Golshah Naghdy, University of WollongongFollow
Fazel Naghdy, University of WollongongFollow
Geoffrey R. Murray, University of WollongongFollow
Haiping Du, University of WollongongFollow

RIS ID

145684

Publication Details

M. Butt, G. Naghdy, F. Naghdy, G. Murray & H. Du, "Patient-Specific Robot-Assisted Stroke Rehabilitation Guided by EEG - A Feasibility Study," in Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, EMBS, 2020, 2020, pp. 2841-2844.

Abstract

© 2020 IEEE. Multi-session robot-assisted stroke rehabilitation program requires patients to perform repetitive tasks. It is challenging for the patient to maintain concentration during training sessions. A novel intervention strategy using Electroencephalography (EEG) signals is proposed to maintain concentration during training by enhancing the engagement of stroke patients using robot-assisted multi-session rehabilitation. The approach is illustrated by applying it to one stroke patient undergoing 12 training sessions of hand motor training on the AMADEO rehabilitation device. AMADEO offers four modes of training programs of increased intensity comprising passive training, passive training with biofeedback, assistive training as well as active 2D training games. The EEG signals are measured over eight electrode sites: FC4, C4, CP4, FC3, C3, CP3, Cz, and CPz during each training day to extract movement-related cortical potential (MRCP) signals. Moreover, functional hand recovery parameters are determined using the AMADEO assessment tool. The patient's level of engagement is determined by the negative amplitude of the MRCP signal. The rehabilitation program is switched to a more intense training mode when a consistent decrease is observed in the negative amplitude of MRCP signals from the monitored electrodes. Using this approach, the rehabilitation program becomes patient-specific and adaptive. In addition, it is shown that each training mode exhibits a different recovery level of the affected hand and maximum recovery is achieved when MRCP signals indicate that the patient is actively participating in the training.