WSEAS Transactions on Biology and Biomedicine
Print ISSN: 1109-9518, E-ISSN: 2224-2902
Volume 14, 2017
Intensive and Repetitive Training with Patient Active Participation though EMG-Controlled Robotic Hand Rehabilitation Device: Healthy Controls and Patients Validation
Authors: , , , , , , ,
Abstract: The objective of this work is to describe and test a hand rehabilitation device with particular attention to the key ingredients for a successful neuro-motor rehabilitation training, and in particular: i) adjunctive high duration and intensity therapy sessions; ii) functional orientation of the training; and iii) patient active involvement. The developed system is composed by a PC, the Gloreha hand rehabilitation glove along with its dedicated screen for visual feedback during movements execution, and the MYO armband for EMG signals recording. Two different control approaches have been designed and implemented taking into account the residual muscular activity of the users: EMG trigger controller, and EMG task-selection classifier. Multiple degrees-of-freedom hand functional movements were alternatively triggered (i.e., when the EMG activity overcomes a predefined threshold, the hand robotic rehabilitation device supports the patient-triggered task) or predicted (i.e., two cascaded artificial neural networks were exploited to detect the patient’s motion intention from the EMG signal window starting from the electrical activity onset up to the movement onset) depending on the selected approach by means of surface EMG signals. The proposed control approaches were tested on nine healthy control subjects (7 females; age range 16-93 years) and a pilot group of four chronic post-stroke patients. All participants, both from the control group and the patients pilot group successfully calibrated and triggered Gloreha during the testing session using the EMG trigger controller. The EMG task-selection classifier demonstrated an overall mean ± SD testing performance of 80% ± 13% and 67% ± 16%. for correctly predicting healthy users’ and pilot post-stroke patient motion respectively. In the control group, the classifier performance was negatively correlated with age, and the pilot patient behaved similarly to elder participants.
Search Articles
Keywords: Electromyography (EMG), EMG controller, artificial neural networks, hand rehabilitation, movement prediction, electromechanical delay
Pages: 29-37
WSEAS Transactions on Biology and Biomedicine, ISSN / E-ISSN: 1109-9518 / 2224-2902, Volume 14, 2017, Art. #5